EEFECT OF REFLOW VARIABLES ON THE NANOSCALE MECHANICAL PROPERTIES OF LEAD FREE SOLDERS

by

TONYE ADEOGBA

Presented to the Faculty of the Graduate School of The University of Texas at Arlington in Partial Fulfillment of the Requirements for the Degree of

MASTER OF SCIENCE IN MATERIALS SCIENCE AND ENGINEERING

THE UNIVERSITY OF TEXAS AT ARLINGTON December 2009

Copyright by Tonye Adeogba 2009 All Rights Reserved

ACKNOWLEDGEMENTS First and foremost, I would like to thank Dr. Pranesh Aswath for his invaluable mentorship. Without his support and dedication, this work could not have been completed. I also thank the members of my committee; Dr. Pranesh Aswath, Dr. Choong-un Kim and Dr. R. Goolsby, for being the core group of faculty in the defense of my thesis, and being great and inspiring educators. Thanks to members of my group who were great friends and supporters to me; Dr. Hande Demikiran, Dr. Bohoon Kim, Arunya Suresh, Mihir Patel, and Hansika Parekh. I cannot forget Jonathan Rowley who I did not personally meet but whose work was followed by me, giving necessary insight for the completion of this study. Further thanks to my good friend Ashley Alcorn for her support A special thanks to Christen and Joyce Adeogba for their unyielding support and encouragement throughout this whole project which could not have been accomplished without them.

December 9, 2009

iii

ABSTRACT

EFFECT OF REFLOW VARIABLES ON THE NANOSCALE MECHANICAL PROPERTIES OF LEAD FREE SOLDERS Tonye Adeogba, M.S.

The University of Texas at Arlington, 2009

Supervising Professor: Pranesh Aswath This study presents findings on the nano-mechanical properties of binary and ternary (Cu, Ag) Sn Solders. Nanoindentation with a cube corner tip was used on Sn-Ag and

SAC405 alloy compositions to investigate the effect of reflow process variables on the properties of hardness and reduced modulus on a nanoscale. The specimens tested were sandwich solder joints where Cu was the substrate. While the reflow variables of cooling rate and reflow time were found to cause changes in IMC morphology and solder matrix microstructure, the nano hardness and reduced modulus were independent of these variables. Low max loads coupled with a cube corner indenter tip yielded interesting load-depth curve results on the solder matrix.

iv

TABLE OF CONTENTS ACKNOWLEDGEMENTS ................................................................................................................iii ABSTRACT ..................................................................................................................................... iv LIST OF ILLUSTRATIONS............................................................................................................. viii LIST OF TABLES ............................................................................................................................xii Chapter Page 1. INTRODUCTION...... ..................................... 1 1.1 Soldering in Electronics Packaging Industry .................................................... 1 1.2 Issues ............................................................................................................... 2 1.3 Nanoindentation as an Important Tool in Electronics Packaging Soldering .... 2 1.4 Research Scope ............................................................................................... 3 2. LEAD FREE SOLDERS AND NANONDENTATION ..................................................... 5 2.1 Solder Requirements and Rationale for (Copper, Silver) Tin as a Lead Free Alternative ............................................................................ 5 2.1.1 Binary Alloys .................................................................................... 6 2.1.2 Ternary alloys ................................................................................... 8 2.2 (Copper, Silver) Tin Intermetallics/Alloys ......................................................... 8 2.3 Reflow Soldering Process of Lead-Free (Copper, Silver) Tin Alloys ............. 12 2.3.1 Reflow Profile ................................................................................. 12 2.4 A Review of Process Variables Influence on (Copper, Silver) Tin Solders ... 14 2.4.1 Microstructure ................................................................................ 14 2.4.2 Mechanical Properties.................................................................... 17 2.5 Nanoindentation Theory ................................................................................ 20 2.5.1 Indenter Types ............................................................................... 21

2.5.2 Hardness and Elastic Modulus Measurements.............................. 22 2.6 Nanomechanical Characterization of (Copper, Silver) Tin Solders/Intermetallics ................................................... 25 2.6.1 Hardness and Reduced Modulus ................................................... 25 2.6.2 Youngs Modulus and Yield Stress ................................................ 26 2.6.3 Creep Rate ..................................................................................... 28 2.6.4 Nanoindentation of Multi-phased Solders ...................................... 30 2.6.5 Indentation Imaging........................................................................ 31 2.7 Objective and Methodology .......................................................................... 34 3. SAMPLE PREPARATION AND EXPERIMENTAL PROCEDURE .............................. 36 3.1 Introduction..................................................................................................... 36 3.2 Solder Joint Tester ......................................................................................... 36 3.3 Stud Preparation ............................................................................................ 38 3.4 Solder Paste Composition .............................................................................. 39 3.5 Sample Preparation ....................................................................................... 40 3.5.1 Profile Control (Intervals) ............................................................... 43 3.6 Sample Analysis ............................................................................................. 44 3.6.1 Polishing ......................................................................................... 44 3.6.2 Nanoindentation ............................................................................. 45 3.6.3 SEM & EDX .................................................................................... 46 3.7 Experiment Objectives and Design ................................................................ 46 4. NANOSCALE PROPERTIES OF PURE INTERMETALLIC Cu6Sn5 & Ag3Sn ......................................................................................................... 49 4.1 Introduction..................................................................................................... 49 4.2 Pure Cu6Sn5 Nanoindentation ........................................................................ 49 4.3 Pure Ag3Sn Nanoindentation ......................................................................... 51

vi

5. EFFECT OF REFLOW VARIABLES ON NANOSCALE PROPERTIES BINARY 96Sn-3.5Ag .................................................................................................. 55 5.1 Effect of Cooling Rate on the Nanoscale Properties of 96.5Sn3.5Ag ........................................................................... 56 5.1.1 Results for Forced Air Cooled (9.97C/s) 96.5Sn3.5Ag ................. 56 5.1.2 Analysis of Forced Air Cooled 96.5Sn-3.5Ag................................. 66 5.1.3 Results for Furnace Cooled (0.38C/s) 96.5Sn3.5Ag .................... 71 5.1.4 Analysis of Furnace Cooled 96.5Sn3.5Ag ..................................... 76 5.1.5 Discussion on the Effect of Cooling Rate on 96.5Sn3.5Ag............ 76 5.2 Effect of Reflow Time on the Nanoscale Properties of 96.5Sn3.5Ag ............................................................................ 82 5.2.1 Results for 96.5Sn3.5Ag Reflowed at 260C for 1s ....................... 83 5.2.2 Results for 96.5Sn3.5Ag Reflowed at 260C for 300s ................... 87 5.2.3 Discussion on the Effect of Reflow Time on 96.5Sn-3.5Ag ........... 91 6. EFFECT OF COOLING RATE ON THE NANOSCALE PROPERTIES OF 95.5Sn4Ag0.5Cu (SAC405) ................................................................................ 92 6.1 Effect of Cooling Rate on the Nanoscale Properties of SAC405 ................... 93 6.1.1 Results and Analysis for Forced Air Cooled SAC405 .................... 93 6.1.2 Results and Analysis for Furnace Cooled SAC405 ....................... 98 6.1.3 Discussion on the Effect of Cooling Rate on the Nanoscale Properties of SAC405 ............................................. 102 7. SUMMARY & CONCLUSION ..................................................................................... 104 APPENDIX A. NANOINDENTATION DATA RESULTS FOR ALL SOLDER JOINT TEST SAMPLES ...................................................................................................... 106 REFERENCES ............................................................................................................................. 117 BIOGRAPHICAL INFORMATION ................................................................................................ 119

vii

LIST OF ILLUSTRATIONS Figure Page

2.1 Compatibility of alloys with reflow soldering ............................................................................... 6 2.2 Phase diagrams of (a) Sn-Ag (b) Sn-Cu .................................................................................. 11 2.3 Phase diagrams of (a) Sn-Ag-Cu (b) Sn rich portion of Sn-Ag-Cu .......................................... 12 2.4 Typical reflow profile................................................................................................................. 13 2.5 SEM photographs of R.C. samples; (a) Sn-3.0Ag-0.5Cu (b) Sn-3.5-0.7Cu and (c) Sn-3.9Ag-0.6Cu of M.C. samples; (d) Sn-3.0Ag-0.5Cu (e) Sn-3.5Ag-0.7Cu, and (f) Sn-3.9Ag-0.6Cu, and of S.C. samples; (g) Sn-3.0Ag-0.5Cu (h) Sn-3.5Ag-0.7Cu and (i) Sn-3.8Ag-0.6Cu ........................................................................... 15 2.6 XRD patters from Sn-3.9Ag-0.6Cu cooled at (a) 0.012C/s and (b) 8.3C/s ........................... 16 2.7 Sn-3.5Ag solder joint (Cu pad) cooled at 0.012C/s ................................................................ 17 2.8 Engineering stress-strain curves in tensile tests at different strain rates and cooling speeds for Sn-3.5Ag-0.7Cu ............................................................................... 18 2.9 Variation of the interfacial IMC layer thickness in as-soldered joints with cooling rates and solder compositions (LEFT) and, variation of single lap-shear strength of the as-soldered joints with cooling rates and solder compositions (RIGHT) ............................................................ 19 2.10 (a) A typical load displacement curve and (b) the deformation pattern of an elastic-plastic sample during and after indentation ............................................................ 24 2.11 A plot of load versus depth for 2.0mN maximum load indentations performed on Sn-Ag-Cu solder, and Ag3Sn, Cu6Sn5 (solid curve) and Cu3Sn (dashed curve) intermetallics ..................................................................................... 25 2.12 (a) Typical indentation load-depth curves for eutectic SnAgCu, SnBi and SnPb solder alloys. (b) Double logarithmic plot of indentation creep strain rate vs. hardness during hold segment for SnAgCu, SnBi, SnPb solders at room temperature .......................................................................................................... 29 2.13 Indentations on (a) primary and eutectic Sn and (b) IMC particles ........................................ 30 2.14 (a) Piling up around the indent on primary Sn (b) the height profiles on each side showing the piling up width ............................................................................................. 32

viii

2.15 SEM images of residual indentation made at a constant loading rate of 800N/s up to 50mN and hold 180s on eutectic SnBi (a), eutectic SnAgCu (b), and eutectic SnPb (c) .......................................................................................................... 33 3.1 Photograph of the final Solder Joint Tester apparatus assembly (minus computer) .............. 38 3.2 Typical reflow profile obtained during pilot test ........................................................................ 42 3.3 Illustration of the VEE Pro software input screen used for insertion of sample parameters and interval heating settings required for preparation of the various solder joints ............................................................................................................................. 44 3.4 A typical cross-section of the solder joint ................................................................................. 47 3.5 Schematic of indentation array in the Cu, interface and solder region .................................... 47 3.6 Load function and parameters from software .......................................................................... 48 4.1 Cu6Sn5 Load-Depth Curve (500N max load) ........................................................................ 50 4.2 Cu6Sn5 Load-Depth Curve (200N max load) ........................................................................ 51 4.3 Ag3Sn Load-Depth Curve (500N max load) .......................................................................... 52 4.4 Ag3Sn Load-Depth Curve (500N max load) .......................................................................... 53 4.5 Slope Characterization of 200n max load on (a) Cu6Sn5 and(b) Ag3Sn (m = 5.7 for Cu6Sn5 and m = 2.87 for Ag3Sn) ........................................................................ 54 5.1 Influence of Cooling Rate on the Fracture Strength of 96.5Sn-3.5Ag Solder Joints (mean values with error bars illustrated) .................................................................................. 55 5.2 Group 1 Indents on FA (9.97C/s) Cooled SnAg ..................................................................... 57 5.3 Group 2 Indents on FA Cooled SnAg (a) SEM image (b) Load-Depth Curves........................ 58 5.4 EDX Mapping of FA Cooled SnAg (Group 2 Indents Location) ............................................... 59 5.5 Group 3 Indents on FA Cooled SnAg Solder Matrix (a) SEM (b) Load Depth Curves ........................................................................................... 61

5.6 Group 5 Indents on FA Cooled SnAg Solder Matrix (a) SEM (b) Load Depth Curves ........................................................................................... 62 5.7 EDX Mapping of FA Cooled SnAg (Group 5 Indents Location) ............................................... 63

5.8 Group 6 Indents on FA Cooled SnAg Solder Matrix (a) SEM (b) Load Depth Curves ........................................................................................... 64

ix

5.9 EDX Mapping of FA Cooled SnAg (Group 6 Indents Location) ............................................... 65 5.10 Load Depth Curve Analysis of FA Cooled SnAg Solder Matrix (a) Group 6 Indents: 1, 3, 4, 8 (b) Group 3 Indents: 6, 9 (c) Group 4 Indent 12 (numbers next to triangles are m with units N/nm) ......................... 68 5.11 Load Depth Curve Analysis of FA Cooled SnAg Solder Matrix (Gradient of Sn matrix) ......................................................................................................... 69 5.12 Creep Characteristics of FA cooled SnAg Solder matrix ....................................................... 70 5.13 Furnace Cooled SnAg Joint (a) SEM (b) Load-Depth curves ................................................ 72 5.14 Furnace Cooled SnAg, SEM of Indents across the whole Solder joint .................................. 73 5.15 Group 3 Indents on Furnace Cooled SnAg (a) Group2 Load-Depth Curve and (b) Group 3 Load-Depth Curve .................................... 74 5.16 EDX Mapping of Furnace Cooled SnAg (Group 2 and 3 Indents Location) .......................... 75 5.17 Load-Depth Curve Analysis of Furnace Cooled SnAg Solder Matrix (a) Group 2 Indents (b) Group 3 Indents ............................................................................. 77 5.18 Creep Characteristics of Furnace Cooled SnAg Solder Matrix .............................................. 79 5.19 SEM & EDX mapping of Furnace Cooled SnAg .................................................................... 80 5.20 Hardness vs. Displacement along SnAg Sample for (a) FA cooled (b) Furnace cooled ......................................................................................... 81 5.21 Fracture strength of 96.5Sn3.5Ag solder joints as function of dwell/reflow time at 260C peak reflow ................................................................................. 82 5.22 SEM of SnAg (260C 1s) Solder IMC joint .......................................................................... 83 5.23 Load-Depth Curve Analysis of SnAg (260C 1s) Solder Matrix .......................................... 84 5.24 EDX Mapping of SnAg (260C 1s) Solder Joint .................................................................. 85 5.25 Creep Characteristics of SnAg (260C 1s) Solder Matrix ................................................... 86 5.26 SnAg (260C 300s) (a) SEM of Indents (b) EDX Phase Map across whole joint ................................................. 88 5.27 Load-Depth Curve Analysis of SnAg (260C 300s)solder Matrix ...................................... 89 5.28 EDX Phase Maps of other Regions in SnAg (260C 300s) ............................................... 89 5.29 Creep Characteristics of SnAg (260C 300s) .................................................................... 90

6.1 Influence of Cooling Rate on the Fracture Strength of SAC405 Solder Joints ........................ 92 6.2 SEM of Groups 1 and 2 Indents across FA cooled SAC405 Solder Joint ............................... 94 6.3 SEM image of Group 3 Indents on FA cooled SAC 405 Solder Matrix .................................... 94 6.4 (a) Load-Depth Curves (b) Load-Depth Curve Analysis of FA cooled SAC405 Solder Matrix ........................................................................................ 95 6.5 EDX Mapping of FA cooled SAC405 Solder Joint ................................................................... 96 6.6 Creep characteristics of FA cooled SAC405 solder matrix ...................................................... 97 6.7 SEM image of Furnace cooled SAC405 joint ........................................................................... 98 6.8 Furnace cooled SAC405 (a) Load-Depth Curves (Group 2 Indents) (b) Load-Depth Curve Analysis of Solder Matrix ..................................................................... 99 6.9 EDX Mapping of Furnace Cooled SAC405 Solder Joint ........................................................ 100 6.10 Creep Characteristics of Furnace Cooled SAC405 Solder Matrix ....................................... 101 6.11 EDX Phase Maps of other Regions in Furnace Cooled SAC405 ........................................ 103

xi

LIST OF TABLES

Table

Page

2.1 Measurements of the Indentation Modulus and Hardness for the (Copper, Tin) Silver system ........................................................................................ 26 2.2 Collected measurements for Youngs modulus of Ag3Sn, Cu6Sn5 And Cu3Sn (Chromiks study assumes a Poissons ratio of 0.33 for all three alloys.). ............................. 27 2.3 Hardness for Ag3Sn, Cu6Sn5 and Cu3Sn as measured by nanoindentation. The estimated yield stress is also given, where the Tabor relation was used ......................................................................................... 28 2.4 Mechanical properties of eutectic SnAgCu, SnBi and SnPb solder alloys obtained from nanoindentation by using Oliver and Pharr method ....................................................................................................... 31 3.1 Characteristics of Copper Bolts used in experiment setup ...................................................... 39 3.2 Solder Paste Profiles ................................................................................................................ 40

xii

CHAPTER 1 INTRODUCTION Soldering is the process in which two metals, each having a relatively high melting point, are joined together by means of a third metal or alloy having a relatively low melting point.(1) The solder which is the third metal or alloy in the definition above, actually undergoes a chemical reaction with both base metals that are being joined to form a new alloy.(2) 1.1 Soldering in Electronics Packaging Industry Solders in microelectronic packaging are applied to bond different electrical components to guarantee the electric connection and mechanical integration of electric devices. In the last decade, lead-free solders have attracted more attention worldwide, because Pb in SnPb solders, which has been used in industry for more than a half century, is harmful to both environment and human health. Much effort has been dedicated to seeking lead-free solder alloys with suitable soldering characteristics and mechanical properties.(3) The electronics assembly practice that has gained momentum in industry over the last 30 years has been the surface mounted assembly (SMA). This assembly comprises of a

surface mounted component soldered on to a printed circuit board. SMAs do not use holes to locate component leads prior to soldering as in the dated through-hole method. Devices

mounted onto Printed Circuit Boards (PCBs) in this fashion are called surface mounted devices (SMD). The precess in which SMDs are joined to the PCBs can be divided into two categories: (a) Component/Solder (CS) processes: Those processes which rely on the insertion/onsertion of components prior to application of solder. This is also known as flow or wave soldering (b) Solder/Component (SC) processes: Those processes which rely on the onsertion of components after the application of solder. This is also known as reflow soldering. It is

important to note that the term reflow is somewhat of a misnomer as it implies previously molten

solder is reheated until molten once more. In reality, many cases involve a solid form of solder in a paste mixture for SC soldering. Reflow soldering is the leading method in industry used to join component to PCBs. During reflow, intermetallic compounds (IMC) grow at a solder/metal interface to form such familiar tin-based compounds as Cu6Sn5 or Ni3Sn4. The process variables involved with reflow soldering greatly affect the morphology of the IMCs formed. With solid-state annealing, these compounds can grow and, often, secondary phases form (e.g., AuSn4 when soldering to gold metallizations). In these cases, fracture through the intermetallic layer results in a decreased strength and lifetime of the solder joint.(4) 1.2 Issues The microelectronics industry faces two significant challenges that make increased knowledge and understanding of the mechanical properties of these intermetallics desirable. The size of devices and solder joints continues to decrease, and the density of integrated circuits grows higher. As a result, the intermetallics comprise a larger volume fraction of the entire package. Additionally, usage of lead-free solders has been shown to produce

intermetallic formations in morphologies not commonly observed in Pb-Sn soldering, such as large Ag3Sn needles and plates that may span an entire solder joint. Knowledge of the

mechanical properties of individual phases makes predictions of the mechanical behavior of an overall joint less demanding and understanding the effect of process variables on the small scale mechanical properties also aids this purpose. 1.3 Nanoindentation as an Important Tool in Electronics Packaging Soldering Significant efforts of research have been focused on the mechanical properties of bulk lead-free solder alloys. Howewer, fewer studies on small-scale experimental characterization have been carried out, particularly on solder joints, especially when it comes to the effect of the process variables on the nano-scale properties. Experimental characterization of small sized

materials plays an important role in reliability evaluation of electronic packaging, and therefore, is critical in package design and manufacturing.(5) Furthermore, efforts are being made to perform Finite-Element-Modeling (FEM) of solder joints at the microstructural level.(6) As a prerequisite to finite element modeling, it is imperative to use reliable materials data. In the special case of microelectronic applications, material properties vary considerably from supplier to supplier. Moreover, the material

properties are often time-dependent and will change during use of the microelectronic component or package. For this reason it is useful to develop miniature tests that are easy to perform, and which operate locally and in-situ. One such test is nanoindentation.(7) When also considering recently developed techniques, such as the low energy Cold-Arc procedure used to connect thin plates of various materials, it becomes apparent that quantitative knowledge of material properties of the joining parts and their change due to the joining procedure is important.(8) The novel nano-indentation technology, which can record continuous load-depth history, has obtained growing attention recently. The merit of this technology can hence ensure the study of the mechanical properties in micro/nano scale precisely. 1.4 Research Scope With advent of legislation limiting and eventually prohibiting the use of lead in electronics, coupled with the manufacturing of increasingly smaller scaled electronics like ceramic resistors used in Surface Mount Technology, the need for proper characterization of lead free solder for PCB assemblies is apparent. Bulk properties of the solders and the effect of process variables on them have been characterized and documented, however, these bulk properties are not fully indicative of the parameters that come into play when dealing with micro to nano scale electronic joints. This work entails: (i) the selection of lead free solder

alternatives, (ii) the use of nanoindentation as a tool for the characterization of in-situ lead free solder nano-mechanical properties (indicative of current microelectronics assemblies), (iii) a discussion on the solder joint morphologies as affected by the reflow

process variables with (iv) interpretation and analysis of collected data for the purpose of understanding the effect of the reflow process variables on the nano scale mechanical properties, (v) further clarifying previous results derived.

CHAPTER 2 LEAD FREE SOLDERS AND NANOINDENTATION 2.1 Solder Requirements & Rationale for (Copper, Silver) Tin as a Lead Free Alternative For more than fifty years, printed wiring boards and components had been designed around the behavior of Sn-Pb eutectic solder during circuit board assembly and in use for holding components to the PWBs. In 1999, NEMI set up pass-fail criteria for lead free

alternatives; the alloy must: (i) have a melting point as close to Sn-Pb eutectic as possible, (ii) be eutectic or every close to eutectic, (iii) contain no more than three elements (ternary composition), (iv) avoid using existing patents, if possible (for ease of implementation) (v) have the potential for reliability equal to or better than Sn-Pb eutectic. The preferred solution for lead-free alternatves varies from region to region, and there is a number of alloys considered promising. The most favorable Pb-free solder systems identified by the industry comprise primarily of Sn with Ag, Bi, Cu, Sb, or Zn.(9) These Sn based alloys can be grouped into binary and ternary systems. The binary systems include: Sn-Sb, Sn-Cu, Sn-Ag, Sn-Zn, Sn-Bi, and Sn-I. The ternary systems include: Sn-Ag-Cu and Sn-Ag-Bi. Ten representative and common alloys were chosen and tested for their compatibility with reflow soldering and the results were presented at the 1999 International Symposium on Microelectronics by Huang et al. The compatibility was quantified as the summation of more primary parameters: wetting ability, solder balling performance, solder tack time, shelf life and solder surface appearance. Figure 2.1 shows the results.

Figure 2.1 Compatibility of alloys with reflow soldering.(9)

By and large, the results show that SnAgBi systems were rated the best on a scale of 1 30 with the factors considered. However, there are further practical reasons as to why other alloys in the selection above are more used in industry. The ratioanle is discuused in the following section. 2.1.1 Binary Alloys Sn-5Sb: Although the fatigue of this alloy is better than that of Sn-63Pb, it requires a substantially higher assembly temperature (~290C) to achieve acceptable wtting

characteristics on Copper. Its high melting point, poor wetting characteristics, and toxicity of antimony limit the application of this alloy.

Sn-0.7Cu: This alloy is recommended by NEMI especially for wave soldering application. It is also the lowest cost lead-free (bar-form) solder and has good wetting charcteristics. However, exhibits lower yield and tensile strength than Sn-37Pb due to the high prevalence of bridges (shorts) and rough solder joints. High process temperatures also limit its application for surface mount PCBs. Sn-9Zn: Although this eutectic alloy has a relatively low meltoing point of 199C and higher tensile strength and fatigue resistance than Sn-37Pb, its major disadvantage is its oxidation characteristics. The oxidation of zinc gives the alloy poor wetting characteristics under normal reflow conditions. Sn-52In: This alloy has a low melting temperature of 118C, however, it has less strength than Sn-37Pb. Sn-81In has a melting point range of 210C - 217C and a higher fatigue strength than Sn-37Pb. However, indium alloys are expensive and the availabilty of indium is limited which excludes its use from electronics. Sn-58Bi: This alloy has a melting temperature of about 138C and is used in low temperature applications. With its superior fatigue resistance over conventional lead-based solder, it has been implemented in consumer electronics and telecommunication applications. Due to the fact that bismuth is a by-product of of lead mining with limited availability, it is not advantageous as a lead-free alloy element. Sn-3.5Ag: This alloy posseses and equivalent fatigue to Sn-37Pb in the 0C to 100C range (not form -55C o 125C), and acceptable wetting characteristics on copper coupons. Moreover, its good resistance to filet lifting plus the aforementioned characteristics explains its long history of use in consumer electronics, telecommunications, aerospace applications and the automotive industry. advantage.(10) Hence, availability (from solder suppliers worldwide) is its major

2.1.2 Ternary Alloys Tin-Silver-Bismuth: There are two possible alloys in this family; Sn-3.3Ag-4.7Bi and Sn-

3.5Ag-1Bi. A 1% bismuth alloy has a melting point of 219C ~ 220C; the 4.7% bismuth alloy has a melting range of 210C ~ 215C. These temperatures and their acceptable wetting characteristics make them suitable for SMAs. Both compositions, more so the 1% bismuth have comparable to superior fatigue restistance than the standard eutectic tin-lead alloy. On the other hand, bismuth productions is a direct result of lead mining, and so its environmental issues coupled with its limited availabiltiy encumbers the use of these alloys. Tin-Silver-Copper: These alloys have become the mainstream lead-free alloys for electronics assembly. NEMI, the Internationl Tin Research Insititute (ITRI) and the european Department of Trade and Industry (DTI) have recommended these alloys. These alloys exhibit acceptable wetting characteristics during board assembly only with a properly designed flux. These alloys have gained acceptance from a host of semiconductor manufacturers including, Mototrola, STMicroelectronics. Alloys with compositions within the range Sn-(3.5~4wt%)Ag-(0.5~1)Cu are close enough to eutectic, they have a melting temperature of 216C to 217C, and are in sufficient supply.(10) It is now apparent why the industry has moved to Tin-Silver and Tin-Silver-Copper alloys as choices for lead free solders. These alloys will be the focus of this research and the

subsequent sections will seek to disect the thermodynamics and morphology of these systems. 2.2 (Copper, Silver) Tin Intermetallics/Alloys Solder joints provide electrical continuity and mechanical stability for interconnects in electronic packages.(2) During the reflow solder process, the metal on the substrate or PCB (Cu Pad) comes into contact with and reacts with molten solder to form an interfacial intermetallic layer. Intermetallic compounds (IMCs) directly influence solder joint properties. Intermetallic compound formation can profoundly change the microstructue of solder joints, and

hence their mechanical properties. Intermetallic compunds can form in the bulk solder and at the solder/metallization interfaces of the solder joints.(2) On a Cu substrate, Sn will react with Cu to form Cu6Sn5 and/or Cu3Sn intermetallic layers. The size and morphology of these layers control the mechanical behavior and reliability of the solder Cu joints. For example, a relatively thin intermetallic layer may be beneficial in achieving a strong mechanical and chemical bond between the solder and Cu substrate. At larger thickness, however, the intermetallic often acts as a crack initiation site leading to catastrophic failure and poor toughness of the joint. Intermetallics also form within the solder, for example, in eutectic Sn3.5Ag solder, where the eutectic mixture consists of Sn and Ag3Sn intermetallic. This intermetallic will also affect the mechanical properties of the solder. Thus, a comprehensive knowledge of the mechanical properties of intermetallic phases formed in Sn rich solders is extremely important. Figures 2.2 and 2.3 displays the different phase diagrams involved with the Tin - (Silver, Copper) system. All phase diagrams are courtesy of the National Institude of Standards and Technology (NIST), Material Science and Engineering Laboratory, Metallurgy Division, Solder Data website. (11)

(a)

(b) Figure 2.2 Phase diagrams of (a) Sn-Ag (b) Sn-Cu

10

(a)

(b) Figure 2.3 Phase diagrams of (a) Sn-Ag-Cu (b) Sn rich portion of Sn-Ag-Cu

11

2.3 Reflow Soldering Process of Lead-Free (Copper, Silver) Tin Alloys An understanding of reflow soldering practice will give a fundamental insight to the process variables that influence the solder joint and its consequent properties. This section details the conventions and parameters necessary to understanding reflow soldering As defined previously, reflow soldering process involves remelting solder previously applied to a joint site in the form of a perform (ingot) or paste, in order to create an attachment. No solder is added during the actual reflow.(12) The following subsection describes the

reflowing soldering temperature profile, which must be optimized to get good soldering result 2.3.1 Reflow Profile Two types of reflow profiles are generally used (1) Ramp-Soak-Spike Profile and (2) Ramp-to-Peak Profile. The newer generation lead-free solder pastes are more suited towards Ramp-to-Peak reflow profiles with a relatively short soak time. Todays more efficient

convection ovens with better heat transfer and temperature control have reduced the need for longer soak times. A typical reflow profile can be broken down into four stages: (i) Preheat Stage: The PCB and component are heated up gradually from room temperature to about 170C. The volatile constituents in the flux evaporate. The ramp rate is of prime importance in this stage of the reflow. It is controlled to within 1-3C/sec. Higher ramp rates may cause thermal shock to the PCB and the components. (ii) Soak Stage: Also known as the dry out or pre-flow stage, the temperature is

maintained between 170 to 220C for an extended period, typically about 60 seconds for a Ramp-to-Peak profile and 120 seconds for a Ramp-Soak-spike profile. This soak helps reduce the temperature delta between the large and small thermal mass components on the assembly during reflow. Also, the flux is activated and it cleans the surface oxides on the solder particles, component leads and PCB pads while keeping the metal surfaces from re-oxidizing. Longer soak times could cause the fluxes to breakdown prior to reflow, leading to solderability issues.

12

(iii) Reflow stage: In this stage, the solder particles melt and form a bond between the PCB and component terminations. Typically a peak temperature range of 235 to 250C at the solder joint with duration above liquidus ranging between 45 75 seconds provides adequate time for wetting and formation of a quality solder joint. Longer reflow times result in excessive intermetallic (IMC will be further discussed) formation enabling more joint brittleness. (iv) Cool Down Stage: During this stage, the assembly is cooled down to room temperature. More rapid cooling helps to form a finer grain structure (typically 3 to 4C/sec). Excessive temperature gradients are avoided to avert possible damage to the components and PCB.(13) Figure 2.3 below displays a typical reflow temperature profile.

Figure 2.4 Typical reflow profile (14)

13

2.4 A Review of Process variables influence on (Copper, Silver) Tin Solders As discussed in the previous section on reflow soldering, the important stages of a typical reflow profile that affect the integrity of the solder joints are the reflow stage and the cooling down stage. This implies that the primary variables that need to be focused on in order to produce the best solder joints are the reflow temperature, reflow time (the duration that the joint is kept at reflow temperature), and the cooling rate. Understanding how these process variables influence the resulting microstructure and consequent mechanical properties is essential to any study on lead free solders. 2.4.1 Microstructure For soldering materials that must also give a structural integrity to electronics assemblies, one needs to understand the solidification phenomena in soldering, which can provide useful information to control microstructure. The effect of cooling rate on the

microstructure of lead free (Cu, Ag) tin joints have been investigated. K.S Kim et al explored the effect of cooling speed on the microstructure of Sn3.0wt.%-Ag0.5wt.%Cu (SAC305), Sn 3.5wt.%Ag0.7wt.%Cu (SAC357), and Sn3.9wt.%Ag0.6wt.%Cu (SAC396). All 3 alloys were received as ingots, then re-melted at 300C for 1 hour, and cooled at three different speeds, 0.012 C s1 (designated as S.C., slowly cooled), 0.43 C s1 (M.C., mildly cooled), and 8.3 C s1 (R.C., rapidly cooled). The influences of both cooling speed and alloy composition on ascast microstructures are shown in figure 2.5 (displayed in a table format). In the case of the R.C. samples, -Sn primary grains are surrounded by a fine eutectic network. samples show microstructures similar to those of the R.C. ones. The M.C.

14

Figure 2.5 SEM photographs of R.C. samples; (a) Sn3.0Ag0.5Cu; (b) Sn3.5Ag 0.7Cu; and (c) Sn3.9Ag0.6Cu, of M.C. samples; (d) Sn3.0Ag0.5Cu; (e) Sn3.5Ag0.7Cu; and (f) Sn3.9Ag0.6Cu, and of S.C. samples; (g) Sn3.0Ag0.5Cu, (h) Sn3.5Ag0.7Cu and (i) Sn3.9Ag0.6Cu.(15)

By summarizing the influence of alloying composition and of the cooling speed, it can be said that the eutectic network is coarsened by decreasing the cooling speed and by increasing Ag content. XRD on SAC 397 alloys (R.C and S.C) shown in figure 2.6, reveal small peaks indicating Ag3Sn and Cu6n5 and strong peaks of -Sn phase. The SAC357 and SAC396 samples showed large platelet precipitates when slowly cooled as shown in figure 2.5h and 2.5i. The authors confirm those platelets to be Ag3Sn via EDX.

15

Figure 2.6 XRD patterns from Sn3.9Ag0.6Cu cooled at (a) 0.012C/s.; and (b) 8.3C/s(15) The effect of cooling rate on the growth of the intermetallic compounds has also been studied by San Won Jeong et al.(16) Sn-3.5Ag, Sn-3.0Ag-0.7Cu, Sn-3.0Ag-1.5Cu, Sn-3.7Ag-

0.9Cu, and Sn-6.0Ag-0.5Cu were sandwiched between Cu Pads and reflow soldered at temperature just above 250C and cooled at 16.67C/s (in water) and 0.02C/s (in furnace). Figure 2.7a corresponds to the Sn-3.5Ag sample cooled in water, and figure 2.7b corresponds to Sn-3.5Ag cooled in the furnace. a1 and a2 on both figures denote the upper and lower side of the joints respectively. It is conclusive in this case also that slower cooling rates allow for more formation Ag3Sn. These micro-structural changes with cooling speed consequently imply changes in the mechanical properties of the alloys.

16

(a)

(b) Figure 2.7 Sn-3.5Ag solder joint (Cu pad) cooled at (a) 16.63C/s and (b) 0.012C/s(16)

2.4.2 Mechanical Properties The same work by K.S. et al Kim sited in the previous subsection also investigated the tensile properties of the samples listed in figure 2.4. Figure 2.8 shows the engineering stress strain curves of the Sn3.5Ag0.7Cu alloy at strain rates ranging from 10
4

to 10

for two

cooling speeds: rapidly cooled (R.C 8.3/s) and slowly cooled (S.C. 0.012C/s). The results show plastic yielding at much lower stresses for the slowly cooled sample as compared with the rapidly cooled sample. Further tests to investigate the Ultimate Tensile Stress (UTS) produced results to the same of effect; the rapidly cooled samples possessed the superior mechanical properties.

17

Figure 2.8 Engineering stressstrain curves in tensile tests at different strain rates and cooling speeds for Sn3.5Ag0.7Cu(15)

The previous subsection reviewed that slower cooling speeds yield more IMC formation, which can be shown to be the explanation for poorer mechanical properties. Figure 2.9 shows the shear strength tests done on the different composition (Cu, Ag)-Sn samples fabricated by San Won Jeong as mentioned in the previous subsection. The results of the tests

18

are shown in the same study and they relate the mechanical shear properties to the IMC thickness as shown below.

Figure 2.9 Variation of the interfacial IMC layer thickness in as-soldered joints with cooling rates and solder compositions (LEFT) and, variation of single lapshear strength of the as-soldered joints with cooling rates and solder compositions (RIGHT).(16) Figure 2.9 reveals that for the same solder compositions there is an inverse relationship of shear strength with IMC thickness. As far as reflow time is considered, studies on the effect of aging also reveal that; longer aging times result in an increase in IMC formation and a consequent decrease in mechanical properties. It can be inferred from this that when reflow times are too long and cooling rates to slow, it is a formula for deficient solder joints.

19

(B) NANOMECHANICAL CHARACTERIZATION OF LEAD FREE SOLDER ALLOYS The fabrication of reliable solder joints in lead free microelectronics packaging depends, in considerable measure, on the formation of an intermetallic compound at the interface between the solder and the finishing of plating material such as Cu, Ni, or Au. As

miniaturization trends in the industry continues with no end in sight, several issues arise. Firstly, smaller packages mean smaller joints, and hence, intermetallic compounds will continue to comprise an increased volume fraction in these new solder joints of reduced dimensions. Another consequence is an increase in local operating temperature so that package components are exposed to severe thermo-mechanical conditions. The solder will suffer more damage to the compact structure of microelectronic devices with different materials. So it is necessary to understand the mechanical behavior of the solder alloys for reliability evaluation of electronic devices.(3) This is not a trivial task as imc reaction layers are in the order of a few microns. Also, bulk intermetallics that have been fabricated by casting and annealing

processes contain residual porosity and/ oxides and the intermetallic in the joint may differ in terms of grain size and defects, with bulk.(17) Nanoindentation has gained momentum as the leading modus operandi for probing mechanical behavior of materials at extremely small scales. In addition, its small volume testing eliminates any adverse effects of oxidation and porosity commonly found in bulk properties. Finally, bulk properties are useless for any good mechanical modeling of todays sized solder joints in the current microelectronics packaging industry. Any accurate finite element model of solder joint behavior requires precise and reliable inputs of the physical properties of the solder matrix and imc. Nanoindentation gives the true in-situ properties of these micron sized layers. 2.5 Nanoindentation Theory Indentation testing is a simple method that consists essentially of touching the material of interest whose mechanical properties such as elastic modulus and hardness are unkonwn with another material whose properties are known. The technique has its origins in mohs

20

hardness scale of 1822 in which materials that are able to leave a permanent scratch in another were ranked harder materials, with diamond assigned the maximum value of 10 on the scale. The establishment of brinell, knoop, vickers, and rockwell tests all follow from a refinement of the method of indenting one material with another. Nanoindentation is simply an indentation test in which the length scale of penetration is measured in nanometers (10 m) rather than microns tests. (10 m) or millimeters (10 m), the latter being common in conventional hardness Apart from the displacement scale involved, the distinguishing feature of most
-6 -3 -9

nanoindentation testing is the indirect measurement of the contact area that is, the area of contact between the indenter and the specimen. In conventional indentation tests, the area of contact is calculated from direct measurements of the indentations of the residual impression left in the specimen surface upon the removal of load. In nanoindentation tests, the size of the residual impression is of the order of microns and too small to be conveniently measured directly. Consequently, it is customary to determine the area of contact by measuring the depth of penetration of the indenter into the specimen surface. This, together with the known

geometry of the indenter, provides an indirect measurement of contact area at full load. For this reason, nanoindentation testing is sometimes referred to as depth sensing indentation (dsi).(18) In nanoindentation testing, the depth of penetration beneath the specimen surface is measured as the load is applied to the indenter. The known geometry of the indenter then allows the size of the area of contact to be determined.(18) 2.5.1 Indenter Types There are various types of indenter geometry used for nanoindentation tests; spherical indenter, vickers indenter, knoop indenter, but the more commonly used ones are the berkovich indenter and the cube corner indenter. Hp Berkovich The Berkovich indenter is used routinely for nanoindentation testing because it is more readily fashioned to a sharper point than the four sided vickers geometry, thus ensuring a more

21

prcise control over the indentation process. The mean contact presure is usually determined from a measure of the contact depth hc (figure 2.4), such that the area of contact is given by:
2 = 3 3 2

is the half-angle of indenter tip and for a berkovich, = 65.3. So therefore:

2 = 24.5

Cube corner In some instances, it is desirable to indent a specimen with more of a cutting action, especially when intentional radial and median cracks are required to measure fracture toughness. A cube corner indenter offers a relatively acute semi-angle that can be beneficial in these circumstances. Despite the acuteness of the indenter, it is still posible to perform

indentation testing in the normal manner and area of contact is the same as that for a berkovich indenter where in this case = 35.26:(18)
2 = 2.597

2.5.2 Hardness and Elastic Modulus Measurements The two mechanical properties measured most frequently using indentation techniques are the hardness, H, and the elastic modulus, E. As the indenter is pressed into the sample, both elastic and plastic deformation occurs, which results in the formation of a hardness impression conforming to the shape of the indenter. During indenter withdrawal, only the elastic portion of the displacement is recovered, which facilitates the use of an elastic solution in modeling the contact process. Figure 2.10 shows a typical loaddisplacement curve and the deformation pattern of an elasticplastic sample during and after indentation. In figure 2.4, hmax represents the displacement at the peak load, pmax. Hc is the contact depth and is defined as the depth of the indenter in contact with the sample under load. Hf is the final displacement after complete unloading. S is the initial unloading contact stiffness.

22

Nanoindentation hardness is defined as the indentation load divided by the projected contact area of the indentation. It is the mean pressure that a material can support under load. From the load displacement curve, hardness can be obtained at the peak load as: =

Where a is the projected contact area. The elastic modulus of the indented sample can be inferred from the initial unloading contact stiffness, s=dp/dh, i.e., the slope of the initial portion of the unloading curve. Based on relationships developed by sneddon for the

indentation of an elastic half space by any punch that can be described as a solid of revolution of a smooth function, a geometry-independent relation involving contact stiffness, contact area, and elastic modulus can be derived as follows: = 2

Where is a constant that depends on the geometry of the indenter ( = 1.034 for a berkovich and cube corner indenter) and er is the reduced elastic modulus, which accounts for the fact that elastic deformation occurs in both the sample and the indenter. Er is given by = 1 2 1 2 +

Where e and v are the elastic modulus and poissons ratio for the sample, respectively, and Ei and vi are the same quantities for the indenter.(19)

23

Figure 2.10 (a) A typical load displacement curve and (b) the deformation pattern of an elasticplastic sample during and after indentation(19)

24

2.6 Nanomechanical Characterization of (Copper, Silver) Tin Solders/Intermetallics 2.6.1 Hardness and Reduced Modulus Nanoindentation has been used in previous literature to characterize the mechanical properties of lead free (copper, tin) silver intermetallics. In these studies, nanoindentation has been used to investigate fundamental mechanical properties such as hardness and modulus of both the solder and intermetallic. One such study can be found in the work of R.R Chromik et al where the hardness and indentation/modulus of Cu6Sn5, Cu3Sn and Ag3Sn were measured. The intermetallic samples were prepared by soldering tin or commercially available solder paste (Sn-4.0Ag-0.5Cu) to either silver or copper substrates. The diffusion couples were subsequently annealed at 200C for times such that at least 5 m of intermetallic would form. Figure 2.11 shows the typical load-displacement data obtained for 2mN maximum load on both the Sn-AgCu solder and the three intermetallics. For a test of the same maximum load, the maximum penetration of the indenter for Sn-Ag-Cu solder is approximately 4.5 times that measured for Cu6Sn5 intermetallic.

Figure 2.11 A plot of load versus depth for 2.0mN maximum load indentations performed on SnAg-Cu solder, and Ag3Sn, Cu6Sn5 (solid curve) and Cu3Sn (dashed curve) intemetallics.(6)

25

The solder was found to be very soft, with a hardness of 0.16 GPa, and exhibited significant plasticity. Upon unloading, the solder recovers only approximately 10 nm of the 460 nm that the indenter penetrated. In contrast to the solder, all three of the intermetallics are found to be significantly harder: Cu6Sn5 (6.5 GPa), Cu3Sn (6.2 GPa), and Ag3Sn (2.9 GPa). The

intermetallics typically recover around 30% of the maximum penetration of the indenter upon unloading. Table 2.1 summarizes the results. From this, the deformation of the intermetallic phases was found to be both elastic and plastic, while the deformation of the solder was found to be primarily plastic.

Table 2.1 Measurements of the Indentation Modulus and Hardness for the (Copper, Tin) Silver system(6)

This study illuminates the use of nanoindentaion in determining basic mechanical properties of intermetallics. Furthermore, the curves derived from the indentation tests also give insights to the deformation behavior of the different phases in a qualitative and quantitative manner. 2.6.2 Youngs Modulus and Yield Stress Other than the primary parameters of hardness and indentation modulus, derivative parameters can also be found from nanoindentationn data; such as the Youngs modulus which requires knowledge of the test materials Poissons ratio, and yield stress y. Another work of R.R Chromik assumes a Poissons ratio of 0.33 to calculate the Youngs modulus of the three intermetallics tested in the previous work above. Table 2.2 shows the results as compared to other Youngs modulus studies of the same materials. While there are discrepancies in

26

agreement, Chromik attributes the non-concensus to lack of knowledge of resulting chemical composition of Subrahmanyan and Cabarets resulting samples. That being said, the results of Ostrovskaya and Fields agree with Chromik that the modulus of Cu 3Sn is greater than that of Cu6n5.(4) Studies done by X.Deng et al have also used nanoindentaion to derive Youngs modulus.(20) Table 2.2 Collected measurements for Youngs modulus of Ag3Sn, Cu6Sn5, and Cu3Sn. (Chromiks study assumes a Poissons ratio of 0.33 for all three alloys.)(4)

R.R Chromik used the Tabor relation to evaluate the yield stress: = Where b is a constant, most often found to be 3. Typically, the Tabor relation extends the usefulness of a hardness measurement because an estimate of the yield stress gives some insight into the materials ability to plastically deform. Their results for the yield stress of the three intermetallics are summarized in table 2.3.

27

Table 2.3 Hardness for Ag3Sn, Cu6Sn5, and Cu3Sn as measured by nanoindentation. The estimated yield stress is also given, where the Tabor relation was used.(4)

2.6.3 Creep Rate Nanoindentation technique could be also used in assessing time dependent properties of materials. As a sharp indenter penetrates into the surface of the sample, an instantaneous strain field is formed in the materials underneath the indenter. The rate at which the

elastic/plastic boundary of the strain field proceeds into the material is thought to determine the indentation creep rate. The strain rate of indentation creep is calculated by taking the time derivative of the displacement during the hold period dividing the instantaneous displacement at that moment. This work is evident in C.Z. Lius study of eutectic SnAgCu, eutectic SnBi and eutectic SnPb solders using nanoindentation.(5) All the indentations were made at a constant loading rate of 800N/s up to a peak load of 50mN. The load ramp was immediately followed by a hold period of 180 s under the peak load so that solder could be subjected to a sufficient creep. Figure 2.12a shows the load-depth curves for the three alloys and figure 2.12b shows the double logarithmic plot of room temperature indentation creep strain rate versus the hardness data from the hold segment of indentation. Both lead-free solders appear more creep resistant than eutectic SnPb solder, with typical hold displacements of about 330 nm for eutectic SnBi solder and of about 440 nm for eutectic SnAgCu solder, respectively. Eutectic SnBi solder shows more creep resistance than eutectic SnAgCu solder. Eutectic SnPb solder is the least creep resistant, with a typical hold displacement of about 670 nm, almost twice as that of eutectic SnBi solder. In figure 2.12b, each curve represents the typical behavior for each

28

solder. The indentation hardness refers to the mean pressure underneath the indenter at that moment, which is found to decrease continuously during the hold segment.

(a)

(b) Figure 2.12 (a) Typical indentation loaddepth curves for eutectic SnAgCu, SnBi and SnPb solder alloys. (b) Double logarithmic plot of indentation creep strain rate vs. hardness during hold segment for eutectic SnAgCu, SnBi, SnPb solders at room temperature.(5)

29

Physical-mathematical derivation of the strain rate during the creep deformation segment can be found in the work of F.Gao et al in their study of the mechanical properties of binary Sn3.5Ag using nanoindentation (21) 2.6.4 Nanoindentation of Multi-phased Solders When dealing with multi-phased solder samples, small nanoindentation loads can be useful in be useful to avoid a single indentation interacting with multiple phases thereby affecting proper data. Yong sun et al used nanoindentation to measure individual phase

mechanical proeperties of lead free solder alloys. Figure 2.13 shows indentations on primary ()sn, eutectic Sn, and Ag3sn intermetallic of the same sac387 (Sn95.5wt%, Ag 3.8wt% and Cu0.7wt%) sample provided by EMC corporation.

Figure 2.13 Indentations on (a) primary and eutectic Sn and (b) IMC particles (22)

30

Different indentation forces were used on Sn and IMCc in this study. 120 n was used for the ag3sn particles and 100n was used for both primary -Sn and Sn in the eutectic mixture. The authors justify the small loads by noting that the size of the IMC particles (or the width of a needle like IMCs) in the as-cast sac alloy was usually around or even under 1m, and that the reasonable flat top surface which is necessary for the indentation is even smaller. Second, in previous studies, the indent depths were in the range of 200500 nm. Also, shallow depths allow for the effect of the indent size/depth on the mechanical properties in nanoindentation to be studied.(22) 2.6.5 Indentaion Imaging After a nanoindentation test, imaging of the indents is also useful in understanding the deformation behavior of the tested solder. For instance figure 2.14 shows piling up around an indent on -sn loaded at 100n. In addition, figure 2.15 also how the size of indents for a given load is informative on different solder. Without knowledge of the hardness values, one can infer from figure 2.9 that eutectic snbi is the hardest of the three alloys due to fact that it has the smallest indent size. This type of qualitative image analysis is useful as a first glance indicator. In fact, the quantitative results of the indents are listed in table 2.4. Table 2.4 Mechanical properties of eutectic SnAgCu, SnBi and SnPb solder alloys obtained from nanoindentation by using Oliver and Pharr method

31

Figure 2.14 (a) Piling up around the indent on primary Sn; (b) the height profiles on each side showing the piling up width(22)

32

Figure 2.15 SEM images of residual indentation made at a constant loading rate of 800N/s up to 50mN and hold 180s on eutectic SnBi (a), eutectic SnAgCu (b), and eutectic SnPb (c).(5)

33

2.7 Objective and Methodology A review of studies done to investigate the effect of reflow process variables on the mechanical properties as related to the microstructure highlight the fact that they were more focused on macro-mechanical properties such as ultimate tensile stresses, and shear strengths. Nano-mechanical characterization on lead free solders using nanoindentaion has been studied. There is ample nanoindentation data on imc, and even though there is less work on lead free solder matrices, it can be found nonetheless. Relating the process variables to the nanomechanical properties however has not been thoroughly investigated. First, it is important to note that this work is following the work of Jonathan Rowley who investigated the effect of process variables on the mechanical properties of binary and ternary lead free solders. The process variables include: the cooling rate and reflow time where an experimental apparatus that mimics the reflow soldering process for a small scale, individual solder joint rather than an entire assembly, was developed the results of that study were summarized in terms of fracture strengths of the joints reflowed. These are predominantly macro-mechanical properties. This study seeks to investigate the effect of the process variables of cooling and reflow time on the nano-mechanical properties of lead free binary and ternary (Cu, Ag) Sn solder intermetallics and matrix. Working with the a set of samples with the same process conditions used in Rowleys work gives the advantage of possibly explaining the macro-mechanical results derived in his work from an in-situ nano-mechanical perspective. The following tasks and goals list the methodology to achieve this goal. (1) Nano-mechanical characterization of pure intermetallics is done as to give a base line for comparison of the in-situ intermetallics. (2) Nanoindentation is used as a tool for investigating the in-situ nano-mechanical properties of the solder joints formed so as to develop a hardness profile from substrate to IMC to solder.

34

(3) The nano-mechanical data is also complimented with scanning electron microscopy for imaging, and energy dispersive spectroscopy as a tool of understanding the indent location composition. (4) Analysis of the load-depth curve should give insight to the deformation behavior of the layers. (5) The compiled results of the substrate, IMC and solder are then used to further clarify the previous macro-mechanical studies done by Rowley on the samples of the same composition

35

CHAPTER 3 SAMPLE PREPARATION AND EXPERIMENTAL PROCEDURE 3.1 Introduction The motivation behind this work meant that it was essential to replicate the reflow soldering process used in electronic packaging assemblies as closely as possible. Replication of such a process was necessary to qualify that the morphology found in an individual solder joint in this work is similar to that formed in electronic assemblies from industry. Process

parameters had to be controllable. By being able to control and vary the parameters, the role each plays could be analyzed. Thus, the process could be optimized and properties of individual lead-free solder joints could be documented. The sample preparation for reflow was executed by Jonathan Rowley for his mechanical properties characterization of the same sample compositions used in this thesis. Hence the sample preparation described in the ensuing

subsections (3.2 3.5) is the work and excerpted from the thesis of Jonathan Rowley.(23) 3.2 Solder Joint Tester A solder joint tester was constructed as the result of an undergraduate senior capstone design project. Modifications were made to the initial layout to further facilitate testing, a The tester is

photograph of the final product and all components is found in figure 3.1.

comprised of a computer (software that controls the heating), adjustable height heating coil, ceramic sheath to encase the heating coil (furnace), air jets for cooling, a two channel thermometer with type K thermocouples, and an adjustable sample stage for joint thickness variability. The computer controlled the process. The software (using VEE Development software from Agilent Technologies) was written so that the desired parameters could be input. The

36

program relayed signals to the heating unit dictating when to cycle on and off, in turn allowing for controlled heating rates and allowing for a consistent temperature for a set hold time. The heating unit (furnace) was comprised of a spiral heating coil one inch in diameter, two inch height encased in a ceramic sheath and covered on the top and bottom with ceramic insulator foam. A hole was cut into the foam to allow the furnace to be adjusted around the test sample. The power supply that activated the heating unit received a signal for the software via a relay (omegabus). The relay received signals from the thermometer then sent then to the computer back to the relay which directed the power supply to remain on or turn off (see appendix 3 for wiring diagram). Thermal energy was applied as long as power was supplied. Thermal energy was cycled in this manor to maintain a semi-constant peak temperature for a set dwell time. The position of the heating coil was vital to the consistency of the heating profile. Positioning of the furnace slightly above or below a certain point in relation to the joint area caused major fluctuations in heating rate and difficulty in the control of the temperature about the set peak temperature. Slower or higher heating rates occurred, and up to 15C over shoots of the peak temperature were observed. The optimum furnace position was detected and

maintained throughout all of the experiments. The sample stage was an adjustable screw type. A clockwise rotation of the shaft would raise the sample stage. A counterclockwise rotation would lower the stage. Both

movements allowed for variations in joint thicknesses. The thickness was monitored by a LVDT (linear variable displacement transducer) attached to the underbelly of the sample stage and cross referenced by actual measurement of cross section thickness. The air jets were employed to allow variability of the cooling rates at the conclusion of the reflow period. The cooling rate was controlled by the quantity of air pressure allowed through the jets. Two adjustable ColeParmer flow meters, capable of applying 14 - 140 LPM of air, were used to control the flow rate. Higher air flow rates yielded higher cooling rates.

37

Other cooling methods were examined to study the effect of cooling rate on the strength as well. Liquid nitrogen was poured directly over the joint for a rapidly cooled specimen. The

furnace was left surrounding the sample to give a very slow cooling rate. Cooling rates were calculated as the average times required to cool from the peak temperature to the melting temperature (time to get from 240-260C to 221 or 217C).

Figure 3.1 Photograph of the final Solder Joint Tester apparatus assembly (minus computer).

3.3 Stud Preparation The copper studs (bolts) had to be machined in order to fit the solder tester in the desired manner. A 3/8 hole had to be cored out of 50% of the studs used. They were to act as the bottom substrates for the mock electronic assemblies. The hole allowed a thermocouple to

38

be inserted into the stud head a few tenths of a millimeter from the solder / copper interface. The stud surfaces (heads) were ground smooth by a sequence of standard grinding procedures: starting at a (1) 120 grit grinding belt, to a (2) 180 grit grinding belt, to a (3) 240 grit abrasive wheel, to a (4) 320 grit abrasive wheel, to a (5) 400 grit abrasive wheel, finishing at a (6) 600 grit abrasive wheel. They were then polished through a series of aluminum oxide polishing

powders: 20m powder, to 5 m, to 1m, to 0.3 m. The studs were washed with soap and exposed to ultrasonic agitation in acetone or isopropyl alcohol immediately before testing. The characteristics of the copper studs used in the experiments are listed in table 3.1. Table 3.1 Characteristics of Copper Bolts used in experimental setup Head Surface Area

Maker

Type

Composition

Length

Threads

Hardness

Fastenal

Hexagonal Head Cap Screw

Cu Si

-20

108.5 2 mm

55.5 HRA

3.4 Solder Paste Composition Each of the two lead-free alloy systems investigated has its own unique composition. Variations in type and content require variations in other additives such as the amount of flux and solvents present. Table 3.2 provides details of the pastes used in this work.

39

Table 3.2 Solder Pastes Profiles Solder Alloy composition Company Name Type of flux Activity level Solvent vehicle 96.5Sn3.5Ag Alpha Metals Omnix OM-310 95.5Sn4.0Ag0.5Cu (SAC405) Alpha Metals Omnix OM-310

Modified Rosin 4-6% w/w ROL-0 No Clean (solvent/semiaqueous) 88% Type 3 (25-45m)

Modified Rosin 4-6% w/w ROL-0 No Clean (solvent/semiaqueous) 88% Type 3 (25-45m)

Metal content Particle size

Rheology Viscosity

Fine Pitch 88% metal M13

Fine Pitch 88% metal M13

3.5 Sample Preparation Obtaining the direct temperature of the sandwiched solder during the testing process was initially challenging. The integrity and legitimacy of the results of the solder joint would be compromised if a foreign intruder or preexisting void were to be present prior to tensile testing. The only method of obtaining the direct temperature of the solder would be to have a thermocouple embedded in the paste while testing. Since a thermocouple could not be placed directly into the solder paste during processing, the idea of obtaining an equivalent temperature was entertained. The idea was to run pilot tests using two thermocouples, one running through the cored hole in the copper stud and one embedded in the paste, to determine the temperature difference between the two locations. It was well known that the temperatures would be However, the

different as the locations of the thermocouple are different in the two cases.

relation/correlation between the two temperatures is important for control of the system.

40

Multiple pilot tests had to be run in order to accurately determine the difference. Knowing the difference of the two, the testing parameters could be set for the stud which would yield the desired temperature in the paste. The recorded temperatures would be that of the stud and not the actual paste. Figure 3.2 illustrates a typical profile of the pilot test; the solid line represents the temperature in the paste and the dashed line represents the temperature of the copper stud. The pilot tests aided in determining cutoff settings to avoid excessive temperature overshoots at the peak reflow temperature. The software inputs allowed for a max. and min. around the peak reflow temperature to maintain a semi-constant peak temperature. The pilot tests aided in determining those settings also. The sample staging area consisted of two parts. The upper stage was adjustable to allow for sample placement, sample removal, and solder thickness control. The lower stage was fixed with a thermocouple protruding upward through its center. The bottom stud was placed over the thermocouple which was adjusted to touch the top of the drilled hole, which put it immediately under the contact surface. Once the bottom stud was in place, the upper stud was lowered until slight contact was made between the two. The studs were aligned and the LVDT was tarred. Prior to this action, the studs were sandwiched together by hand and the combined stud head thickness at all six sides was determined via calipers. The average of the six measurements was recorded as the stud thickness. After alignment and tarring, the top stud was raised and the bottom stud was placed on a scale where the solder paste was applied by syringe directly onto the stud surface and the weight recorded. The stud was then repositioned on its stage and the top stud was lowered. The top stud was lowered until the LVDT read a desired paste thickness and uniformly spread the paste over the entire stud surface. Once achieved, thermal energy was applied to reflow the solder. The computer program was activated and testing parameters were input prior to heat activation. The program required the input of the following information: 1) sample name, 2) operator, 3) current room temperature, 4) stud thickness, 5) paste thickness (from LVDT), 6)

41

paste weight, 7) interval settings, 8) collection times, and 9) cooling method. Figure 3.5 illustrates several of these inputs. Each lead-free alloy tested required separated settings. Once all parameters were entered, testing (heating) began. The data (time and temperature) was fed into a data file that could be transferring into a spread sheet for plotting of the curves. At the conclusion of the reflow period, the selected cooling method was applied. Cooling continued until the cutoff (steady state) temperature was reached. At that point, data collection ceases and the sample is removed from the solder tester. Upon cooling to near room temperature, the entire assembly (studs plus solidified solder) was measured by caliper on all six sides and averaged. The average was recorded as the assembly thickness. The difference between the assembly thickness and stud thickness (without solder) was the reported joint thickness.

Figure 3.2 Typical Reflow Profile obtained during pilot test

42

3.5.1 Profile Control (Intervals) With replication of the industrial reflow soldering process in mind, the computer software was written so that the heating rates could be varied to correspond with those of the zones used in industry packaging production. Three intervals were written into the program; interval 1 corresponds to the preheat zone, interval 2 to the soak zone, and interval 3 to the reflow zone. Figure 3.5 illustrates the interval inputs screen. Final Temp and Interval Time are the major factors in zone heating rate control. The Final Temp setting corresponds to the temperature at which that interval (zone) ends and the next begins. The Interval time setting corresponds to the time (in seconds) in which it was desired to reach the Final Temp. The heating rate for interval 1 was calculated as the difference between room temperature and Final Temp 1 divided by Interval Time 1. The heating rate of interval 2 was calculated as the difference of Final Temp 2 minus Final Temp 1 divided by Interval Time 2. For interval 3, the Final Temp corresponds to the peak temperature and Interval Time represents the dwell time at peak temperature (dwell zone). Since there were no components or corresponding worry of thermal shock, the soak zone was bypassed during testing. The heating rate was held constant up to the reflow zone. The rate was changed at that point to slow down heating so that the peak temperature would be reach but over shoot minimized.

43

Figure 3.3 Illustration of the VEE Pro software input screen used for insertion of sample parameters and interval heating settings required for preparation of the various solder joints.

3.6 Sample Analysis 3.6.1 Polishing After reflowing and cooling, the samples were mounted using an epoxy resin with 2:1 powder to liquid weight ration provided by LECO. The mounted samples were polished using standard metallographic methods. First, each sample was grinded using a 120 grit silicon carbide paper on a rotating belt to generate unsullied surface cross-section. Further grinding was done using 240 grit, 320 grit, 400 grit and lastly a 600 grit silicon carbide paper to smoothen the surface. The samples were then polished using 20m alumina powder/water mixture on a corresponding micro-cloth provided by LECO. The polishing is repeated using

44

5m, 1m, and finally a 0.3m alumina powder/water mixture on their respective micro-cloths for a mirror finish. 3.6.2 Nanoindentation Nanoindentation tests were carried out on the polished specimens using a Hysitron Inc. triboscan Nanoindnenter accompanied with a Ubi 1 sofwatre version 8.1.1. The experiment was carried out as followed (i) The sample was marked to demarcate where the indentations were to be carried out on the sample. This also aided in identifying indentation locations when Scanning Electron Microscope imaging. (ii) The sample was then mounted on a steel plate so as to be properly placed on the nanoindenter magnetic stage as this ensures stability and zero movement during indentation tests. Figure 3.4 shows a picture of a typical mounted sample.. (iii) The nanoindenter optical live camera was then used to view the sample to align the indenter cube corner tip as close to solder and marked region as possible. (iv) The nanoindenter was then set to automate 13 indentations in a line array from the copper region moving towards to the solder. The 13 indentation spanned a length of 30N (v) To avoid the problem of missing areas, the beginning of the next array was moved +20m from the previous array by moving the nanoindenter stage +20m in the x-direction closer to the solder. Also, to avoid overlapping indentations, the next array was also moved +20m in the ydirection from the previous, and the third array moved -20m in the y-direction. Hence, every other array was in the same y-position. This was continued until the arrayed indentions were well into the solder and past the intermetallic region in the x-direction as shown in the schematic in Figure 3.5. Load Function The load function of each indentation is a trapezoidal function with a total duration of 15 seconds. The loading, rest, and unloading sections were 5 seconds each. The nanoindenter

45

was set to carry out the indentation in load control mode with the peak (rest) load set to 500N. Figure 3.6 shows the software load function. Due to the fact that the solder is much softer and porous than the copper and intermetallic, the nanoindenter could not record the appropriate curves necessary for hardness data at 500N peak load. To avoid this problem, once the solder region was reached, the peak load for the array indentations was set to 200uN. Although larger loads would reveal larger indentations in the harder IMC region more conducive to imaging, a hardness profile from substrate to IMC to solder was a main goal of this research. Hence, large discrepancies in data resulting from indentation size effect had to be avoided. This was another reason why a peak load of 500N, not too far from 200N was chosen to start form the Cu substrate side. 3.6.3 SEM & EDX A Hitcahi S-3000N model Scanning Electron Microscope was used to identify and image the indentations to correspond with the nanoindentation data. Energy Dispersive

Spectroscopy was also used to map out the chemical composition distribution of the Cu-IMCSolder interaction so as to give further insight to the nano-mechanical properties observed. 3.7 Experiment Objectives and Design In order to investigate the effect of cooling rate, both the binary SnAg and ternary SAC405 were examined at the fast the fast Fan/Forced Air cooling rate of 9.97C/sec and the slow Furnace cooling rate of 0.38C/sec. In order to investigate the effect of reflow time, two samples of binary Sn-Ag composition were tested, distinguished by their reflow times of 1 second and 300 seconds at a reflow temperature of 260C

46

Cu

Solder

Figure 3.4 A typical cross-section of the solder joint

Group 1 indents

Group 3 indents

Figure 3.5 Schematic of indentation array in the Cu, interface and solder region

47

Figure 3.6 Load function and parameters from software

48

CHAPTER 4 NANOSCALE PROPETRIES OF PURE INTERMETALLIC Cu6Sn5 & Ag3Sn 4.1 Introduction Before examining the actual solder joint, it was important to determine properties of pure intermetallic compound Cu6Sn5 and Ag3Sn, it was necessary to characterize pure intermetallics so as to gain perspective on the properties of the in-situ intermetallics. To

achieve this, pure intermetallics of Cu6Sn5 and Ag3Sn was obtained, courtesy of Northwestern University. Single phase bulk specimens were prepared by conventional casting followed by annealing for long times. Nanoindentation was performed on both samples where 9 different locations were indented on each sample at 500N max load and repeated at 200N max load. The reason for also testing at 200N max load was to compare IMCs in the solder joint and IMC islands subjected to 200N in the methodology laid out in 3.6.2. The results are promptly discussed below. 4.2 Pure Cu6Sn5 Nanoindentation 500N max load The load depth curves of Cu6Sn5 for a max load of 500N in figure 4.1, show a penetration depth of approximately 80 90nm for the initial loading segment. An average hardness of 6.04 0.5GPa is recorded with an average reduced modulus Er of 122 3GPa. The hold segment of the curves gives insight to the creep resistance properties of the IMC. This dwell segment shows an average depth increase of 5nm. The unloading segment shows that the Cu6Sn5 recovers about 30% of initial loading penetration depth.

49

Load - Depth Curve for Cu6Sn5 Pure Intermetallic (500N max load) 600

500

indent 1 indent 2

400

indent 3 indent 4

Load (N) 300

indent 5 indent 6

200

indent 7 indent 8

100

indent 9

0 0 20 40 60 Depth (nm) Figure 4.1 Cu6Sn5 LoadDepth Curve (500N max load) 200N max load The hardness and reduced modulus results of the 9 indents made on pure Cu 6Sn5 intermetallic at a max load of 200N is of 7.35 0.6GPa with an average Er of 128 4GPa. While Er measured here is in overall agreement with the measurements at 500N, the hardness measurement is 18% higher. One explanation for this slight increase is an effect called the indentation size effects where lower nanoindentation loads result in minor increases in hardness measurements for the same material. One reason is smaller loads leave smaller indents 80 100 120

allowing for more discrepancy between indent area and area function used by the nanoindenter software. There are documented studies for correction of this effect. With all that to say, the range between both tests is satisfactory.

50

Load - Depth Curve for Cu6Sn5 Pure Intermetallic (200N max load) 250

200

indent 1 indent 2

150 Load (uN) 100

indent 4 indent 5 indent 6 indent 7

50

indent 8 indent 9

0 0 10 20 30 40 50 Depth (nm) Figure 4.2 Cu6Sn5 LoadDepth Curve (200N max load) The loading segment of the curves yields an average penetration of 34 41nm for the alloy. The dwell segment gives a depth increase of 2 4nm on average. After unloading, the alloy recovers about 38% of max load segment penetration. (Refer to table 2.1 and 2.3 for comparison of values measured in this subsection with other studies) 4.3 Pure Ag3Sn Nanoindentation 500N max load The load-depth curves of the 500N max load test are shown in figure 4.3 where a loading segment penetration of 140 160nm can be seen.

51

Load - Depth Curve for Ag3Sn pure Intermetallic (500 N max load) 600

500

indent 1 indent 2

400 load (N) 300

indent 3 indent 4 indent 5 indent 6 indent 7

200

indent 8 indent 9

100

0 0 50 100 Depth (nm) 150 200

Figure 4.3 Ag3Sn LoadDepth Curve (500N max load) An average hardness of 2.55 0.2GPa is recorded with an average Er of 86 2GPa. The dwell segment shows 10nm of depth increase for all curves, and 20% recovery from loading segment penetration. 200N max load Figure 4.4 shows the loading curves for the indents on Ag 3Sn at a 200N max load. Indents 1, 5, and 9 were on voids caused by outlying curves so they were removed. The rest are with little deviation. The average hardness recorded here is 2.990.5GPa and the reduced modulus is 93.27GPa. The loading segment yields a penetration of depth between 65 and 75

nm. The dwell segment produces a 6nm depth increase, and the unloading segment allows for a 29% recovery from load segment penetration.

52

Load - Depth Curve for Ag3Sn Pure Intermetallic (200N max load) 250

indent 2 200 indent 3 150 Load (N) indent 6 100 indent 7 indent 8 indent 4

50

0 0 20 40 60 Depth (nm) 80 100

Figure 4.4 Ag3Sn LoadDepth Curve (200N max load) Finally, slopes for the loading segment for the 200N max for both model compounds were evaluated in the hopes of characterizing IMC particles embedded within the solder matrix during experimentation on the test compounds. Figure 4.5 shows the slopes. . (Refer to table 2.1 and 2.3 for comparison of values measured in this subsection with other studies)

53

(a)

(b) Figure 4.5 Slope Characterization of 200N max load on (a) Cu6Sn5 and (b) Ag3Sn (m = 5.7 for Cu6Sn5 and m = 2.87 for Ag3Sn)

54

CHAPTER 5 EFFECT OF REFLOW VARIABLES ON NANOSCALE PROPERTIES OF BINARY 96.5Sn3.5Ag This chapter displays and discusses the results of the mechanical and nano-mechanical properties of binary 96.5Sn-3.5Ag (Sn-Ag) solder and the effect of cooling rate and reflow time on the properties. The macro-mechanical properties were summarized in terms of fracture strengths studied by Jonathan Rowley. Nanoindentation data is then discussed to investigate if there is any effect of cooling rate on the nano-mechanical properties of the binary solder.
INFLUENCE OF COOLING RATE ON 96.5Sn3.5Ag SOLDER FOR 1s AT PEAK REFLOW
100.00 95.00 90.00 85.00 80.00
FRACTURE STRENGTH (MPa)

89.52 85.85 78.78

75.00 70.00 65.00 60.00 55.00 50.00 45.00 40.00 35.00 30.00 25.00 20.00 0 1 2 3 4 5 6 7 8 9 10 11 COOLING RATE (C/sec) 31.46 FURNACE COOL (0.38 C/sec) ROOM AIR COOL (2.17 C/sec) LIQUID NITROGEN COOL (3.32 C/sec) FORCED AIR (50 lpm) COOL (9.97 C/sec)

Figure 5.1 Influence of Cooling Rate on the Fracture Strength of 96.5Sn-3.5Ag Solder Joints (mean values with error bars illustrated)(23)

55

Figure 5.1 shows the fracture strength of Sn-Ag as affected by the cooling rate. There is a clear degradation in fracture strength with decreasing cooling rate. It can also be seen that the slowest cooled sample (0.38C/s) is significantly inferior to the rest of other samples. 5.1 Effect of Cooling Rate on the Nanoscale Properties of 96.5Sn3.5Ag This section is divided into five subsections demarcated by the two different Sn-Ag samples tested. Subsection 5.1.1 5.1.2 corresponds to the Forced Air cooled sample

(9.97C/s), and 5.1.3 5.1.4 corresponds to the Furnace cooled sample (0.38C/s) while 5.1.5 discusses the results. The analysis method used here is used as a standard methodology for the other control groups in this work, including ternary (chapter 6) so a more detailed results presentation and analysis is done here. 5.1.1 Results for Forced Air cooled (9.97C/s) 96.5Sn3.5Ag Recalling the methodology of 3.6.2 and figure 3.5, groups of 13 indents each are performed on the FA cooled Sn-Ag sample. Sn-Ag FA cooled Cu substrate (group 1, 500N max load) All indents in group 1 are on Cu at a max load of 500N and the results gave an average Cu hardness of 2.33GPa with an average reduced modulus (E r) of 118GPa for a 500N max load. Figure 5.2 displays all 13 indents in group1 which can be seen to be even sized, manifesting the homogeneity of the Cu substrate.

56

Figure 5.2: Group 1 Indents on FA (9.97C/s) Cooled SnAg Sn-Ag FA cooled Solder Joint (Group 2, 500N max load) Group 2 indents are loaded at 500N max load starting from Cu and ending up in the solder matrix. There is no data for indents 8, 9, 10, 12, and 13 as these indents are on the solder which is too soft for a max load of 500N to record an appropriate curve. Indent 7 records a hardness of approximately 1GPa and indent 11 records a hardness of 4.1GPa as shown in SEM of group 2 in figure 5.3(a). Figure 5.3(a) shows that indent 7 is slightly larger than indents 1 6 on Cu, but smaller than indents 8, 9, 10, 12, and 13 on the solder. EDX mapping in figure 5.4 (corresponding to SEM figure 5.3a) shows that indent 7 is at the Cu6Sn5 intermetallic joint and the Sn-Ag solder matrix interface. It suggest a very shallow intermetallic layer above solder because it is not as hard as indent 11 location which can be seen in figure 5.4 to be an island of Cu6Sn5 intermetallic.

57

(a) LOAD - DEPTH CURVE of GROUP 2 indents on Sn-Ag (FA cooled 9.97C/s) 600 500 400 Load (N) 300 200 100 0 0 50 100 150 200 Depth (nm) (b) Figure 5.3 Group 2 Indents on FA Cooled SnAg (a) SEM image (b) Load-Depth curves 250 300

indent 11

indent 7

indent 1 indent 2 indent 3 indent 4 indent 5 indent 6 indent 7 indent 11

58

Cu6Sn5 island (indent 11)

Figure 5.4 EDX Mapping of FA Cooled SnAg (Group 2 Indents Location)

59

Sn-Ag FA cooled Solder matrix (Groups 3 and 4, 200 max load) Groups 3 and 4 indents (200N max load) are all in solder matrix region. The average solder hardness taken from the two groups is approximately 0.320.05GPa. The average

solder modulus taken from group 3 is 60.65.5GPa. Figure 5.5 shows the micrograph and loaddepth curves of group 3 indents in the solder matrix. For clarity, the anomalous curves resulting from indents on voided or deformed surface morphologies are omitted. The deformation

behavior is interesting as a wavy profile can be observed from the loading segments of the curve. Analysis of this load segment profile of the solder matrix is discussed in later

subsections. Penetrations from the loading segments of the indents range from 200 to 300 nm. The hold segment at a max load of 200N shows net depth increases of about 50 to 63 nm. From the unloading segment in figures 5.5b, one can observe that all of the curves have negligible recoveries with the exception of indent 1 which shows some elastic recovery when the load is almost completely removed at less than 10N. Sn Ag FA cooled Solder matrix with IMC islands (Groups 5 and 6, 200N max load) Groups 5 and 6 (200N max load) are also in the solder region but possess some indents on islands of Cu6Sn5 IMC (Figures 5.6(a) and 5.8(a)). Group 5 indent 5 and group 6 indents: 3, 4 record hardness values of 4.18, 4.18 and 4.39GPa respectively for the Cu6Sn5 IMC. The average reduced modulus Er the indents is 94.9GPa6GPa. The SEM and load depth curves of group 5 indents can be seen in figure 5.6. Figure 5.7 is the EDX mapping of group 5.

60

(a) LOAD - DEPTH CURVE of Group 3 indents on Sn-Ag (FA cooled 9.97C/s) 250 indent 1 200 indent 2 indent 3 indent 4 indent 5 indent 6 50 indent 9

150 Load (N) 100

0 0 100 200 Depth (nm) (b) 300 400

Figure 5.5 Group 3 Indents on FA cooled SnAg Solder Matrix (a) SEM (b) Load Depth Curves

61

(a) LOAD - DEPTH CURVE of Group 5 indents on Sn-Ag (FA cooled 9.97C/s) 250

indent 6 indent 5 indent 8

indent 2 indent 3 indent 4

200

150 Load (uN) 100

indent 5 indent 6 indent 7 indent 8 indent 10

50

indent 11 indent 12

0 0 100 200 Depth (nm) (b) 300 400

indent 13

Figure 5.6 Group 5 Indents on FA cooled SnAg Solder Matrix (a) SEM (b) Load Depth Curves

62

Figure 5.7 EDX Mapping of FA Cooled SnAg (Group 5 Indents location)

63

Group 6

(a) LOAD - DEPTH CURVE of Group 6 indents on Sn-Ag (FA cooled 9.97C/s) 250 indent 1

200

Indents 3, 4

indent 2 indent 3 indent 4 indent 5 indent 6 indent 7 indent 8 indent 9

150 Load (N) 100

50

indent 6

indent 10 indent 11 indent 12 indent 13

0 0 100 200 Depth (nm) (b) 300 400

Figure 5.8 Group 6 Indents on FA cooled Solder Matrix (a) SEM (b) Load Depth Curves

64

Ag rich precipitates

Figure 5.9 EDX Mapping of FA Cooled SnAg (Group 6 Indents Location)

65

From figure 5.6b and 5.8b, it is observed that all three IMC indents show maximum load segment penetrations (40nm) that are approximately 5 times less than penetrations in the solder. It can also be inferred that the IMC typically recovers around 30% of the maximum load segment penetration of the indenter after unloading. The wavy loading segment in the solder is also visible here. Group 6 - indent 6 shows a huge elastic recovery that is not typical of all other solder indents. The reason for this can be seen in SEM figure 5.8a which reveals a bright tiny blotch on the location. The brighter spots can be seen scattered in the solder matrix as seen in SEM images; figures 5.5a 5.8a. They are Ag3Sn particles in the solder matrix. They are also observed from the Ag spectrum EDX mapping of group 6 (figure5.9). The elasticity of Ag3Sn as compared with the solder induced the recovery of this indent. This is the case for group 6 indent 7, and group 3 - indent1. It is well known that solidification of Sn-Ag in most conditions will yield a multiphase solid of -Sn matrix with Ag3Sn intermetallic precipitate particles and if Cu is present, Cu6Sn5 imbedded in the matrix. With the knowledge, that a cube corner tip is markedly sharp as compared to the more popular Berkovich tip, coupled with the very low load of 200N applied to the solder matrix, sensitivity to the different phases within the matrix could be realized with the nanoindenter during penetration. This sensitivity is manifest in the wavy profile of the loading segment observed during indentation on the solder matrix. Analyzing this loading segment will give insight as what phases the indenter comes into contact with during penetration. 5.1.2 Analysis of Forced Air Cooled 96.5Sn3.5Ag Recalling the characterization of the model intermetallics in Chapter 4, the slopes of the loading segments were evaluated. If it is indeed the case that that the wavy oscillations are due to the different phases present; Ag3Sn and Cu6Sn5 in the -Sn matrix, then the steep gradients in the oscillations, source from either of the IMC particles and the flatter gradients are the

66

loading profiles for Sn. The slope gradients for Cu6Sn5 and Ag3Sn model compounds loaded at a 200N max load is 5.7 and 2.87 respectively. The analysis of the segmented slopes would be to evaluate their gradients and identify which particle is present at that point of the loading segment comparing with the slopes of the model compounds load depth curves (Chapter 4). Consequently, a sharp change in gradient implies a change in contact of phase and the abscissa of a step slope (projected on to the depth axis of the load depth curve) would be the displacement of interaction between the particle and the indenter tip. Figure 5.10 shows the analysis of the FA cooled sample where the oscillated loading segment of select indents on the solder matrix is dissected. Henceforth: Slopes m 3 are identified as Cu6Sn5. Slopes m < 3 are identified as Ag3Sn. Near flat slopes of 0.1 0.3 are Sn matrix. Numbers next to slope triangles in the Load-Depth Curve Analysis figures are the gradients (m) with unit N/nm. The fact that these IMC particles are embedded in a soft matrix is justification for the slopes being somewhat less than the corresponding model IMC. It can be seen in figure 5.10 that there are indeed numerous IMC particles of Ag3Sn and to a lesser degree, Cu6Sn5 in the FA cooled SnAg solder matrix. The analysis shows that larger interactions of Ag3Sn with the nanoindenter tip are in the order of 15nm while the smaller particle interactions are 4 5nm. Cu6Sn5 interactions with indenter tip are less than 10nm in this case.

67

(a)

(b) Figure 5.10 Load Depth Curve Analysis of FA Cooled SnAg Solder Matrix (a) Group 6 indents:1,3,4,8 (b) Group 3 indents: 6,9

68

(c) Figure 5.11 Load Depth Curve Analysis of FA Cooled SnAg Solder Matrix (Gradient of Sn matrix) The flatter slopes evaluated in figure 5.11 characterize the Sn matrix where the projected abscissa of these segments is analogous to the average spacing of the IMC particle interaction with the nanoindenter tip. The analysis in figure 5.10 has clarified and dissected the unique profile involved with the loading segments of indents performed on the solder matrix; identifying IMC nanoscale interactions with the nanoindenter tip. The wavy profiles are universal to the loading segments of solder indents performed on the FA cooled sample as can been in the load depth curves in solder regions. The depth increase of the dwell segment is a measure of the creep properties. While Cu6Sn5 islands characterized in this sample show negligible depth increase at load dwell, the solder matrix does. To characterize the creep properties, a plot of contact area stress vs. creep depth and a plot of creep depth versus initial load segment depth is shown in Figure 5.12 with data from groups 3 and 4 indentations on the solder.

69

The contact Ares stress is =

where Pmax is the max load at dwell (200N) and

A = 2.597hc for a Berkovich tip as defined in section 2.9. Figure 5.12b shows that indents with higher initial loading depths are subject to higher stresses between the indenter and the solder therefore causing less creep. It can be deduced form figure 5.12a the contact area stress (analogous to the normal stress or pressure induced from Sn matrix) is a significant component of the total stress induced on the tip. Contact Area stress vs. Creep depth for FA cooled Sn-Ag 1.1 1 0.9 Contact Area 0.8 stress (Gpa) 0.7 0.6 0.5 40 45 50 55 60 Creep Depth (nm) (a) 65 70 75

Creep depth vs. Initial Load Depth fo FA cooled Sn Ag 80 70 Creep depth (nm) 60 50 40 30 20 200 220 240 260 280 Initial load depth (nm) (b) 300 320 340

Figure 5.12 Creep Characteristics of FA cooled SnAg Solder matrix

70

5.1.3. Results for Furnace Cooled (0.38C/s) 96.5Sn3.5Ag For the furnace cooled sample, only three groups of indents each are performed on the furnace cooled Sn-Ag sample primarily because of the small solder thickness in this sample. Figures 5.13 to 5.16 each show (a) SEM images (b) the load depth curves, and EDX mapping (where necessary) of the indents in groups 1 to 3 respectively. Sn-Ag Furnace cooledSolder Joint (Group1, 200N max load) Group 1 indents are loaded at a max of 200N where most of the indents are on Cu, but indents 11, 12 are on Cu6Sn5 intermetallic. Figure 5.13 shows the corresponding SEM and load-depth curves. The average IMC hardness measured from the two indents is 4.18GPa with an average Er of 96.7GPa. Indent 13 is on the solder. Due to the fact that the max load is at 200N, indents on Cu are barely visible and those on IMC are invisible at 2500K (Figure 5.13a). The loading segment manifests a penetration of 40nm; shorter than Cu 60 to 70nm, and much shorter than solder = 300nm. The creep resistance of the Cu 6Sn5 IMC is slightly greater than that of the Cu substrate as is evident from the dwell segment.

71

(a) LOAD - DEPTH CURVE of Group 1 indents on Sn-Ag (Fur. cooled 3.32C/s) 250 indent 1

indents 11, 12
200

indent 13

indent 2 indent 3 indent 4

150 Load (N) 100

indent 5 indent 6 indent 7 indent 9

50

indent 10 indent 11 indent 12

0 0 100 200 Depth (nm) (b) 300 400

indent 13

Figure 5.13 Furnace Cooled SnAg Joint (a) SEM (b) Load-Depth curves

72

Solder Matrix with IMC islands (Group 2 and Group 3, 200uN max load) In figure 5.14 Group 2 - indents 3, 4 and Group 3 Indents 9 are not visible because they are on the Cu6Sn5 IMC joints where the average hardness recorded is 4.13GPa and average Er is 94.5Gpa. Indents 5, 6 are on a void with a very rough morphology and are also not visible. Group 2, indents 7 9, and group 3, indents 2 - 7are on the solder matrix where an average hardness of 0.3GPa is measured with an average Er of 6310GPa. Group 2, indents 10 - 12 and group 3, indent 1 are on a Ag3Sn platelet which is confirmed in figure 5.16. The three indents on this platelet record an average hardness of 3.04GPa and an average Er of 76GPa. Figure 5.14a shows the load-depth curves of group 2 indents and Figure 5.15b shows the load depth curves of group 3 indents. The hold segment as a measure of creep

resistance is virtually indistinguishable for both IMCs (Ag3Sn and Cu6Sn5) with around 5 8 nm of depth increase. (Figure 5.15)

Figure 5.14 Furnace Cooled SnAg, SEM of Indents across the whole Solder joint

73

LOAD - DEPTH CURVE of Group 2 indents on Sn-Ag (Fur. cooled 0.38C/s) 250

indents 3,
200

indents 10, 11, 12

indent 3 indent 4 indent 8 indent 9 indent 10 indent 11 indent 12 indent 13

150 Load (N) 100

50

0 0 50 100 150 200 250 300 350

Depth (nm) (a) LOAD - DEPTH CURVE of Group 3 indents on Sn-Ag (Fur. cooled 0.38C/s) 250

indent 9
200

indent 1
indent 1 indent 2

150 Load (uN) 100

indent 3 indent 4 indent 5 indent 6

50

indent 7 indent 9

0 0 50 100 150 200 Depth (nm) (b) Figure 5.15 Group3 indents on Furnace Cooled SnAg (a) Group2 Load-Depth curve and (b) Group3 Load-Depth curve 250 300 350 400

74

(b) Figure 5.16 EDX Mapping of Furnace Cooled SnAg (Group 2 and 3 Indents Location)

75

5.1.4 Analysis of Furnace Cooled 96.5Sn3.5Ag Figure 5.17 shows the analysis of the loading segments in the solder region penetration of the furnace cooled sample. Particles of Ag3Sn but more so Cu6Sn5 are contacted by the indenter in the matrix. The slope of the Cu6Sn5 joint and Ag3Sn platelet is found to be within close vicinity to the corresponding model IMCs. This suggests a thick layer of the compounds due to the slow cooling rate allowing for more nucleation of the IMC. The large Ag3Sn platelet explains why the particles of Ag3Sn are less found in the matrix around this region as a large amount has diffused away from the matrix to platelet formation. The creep characteristic of the solder is seen in figure 5.18 (a) plot of area contact stress vs. creep depth and (b) creep depth vs. initial loading depth. The trend that deeper loading depths are subject to less creep is evident but with some scatter. It should be noted that more points on the solder would yield more correlation but the region tested in this sample lacked ample solder area. There is no evidence, however, of a relationship between creep depth and contact area stress. 5.1.5 Discussion on the Effect of Cooling Rate on 96.5Sn0.5Ag The results show that slower cooling does not seem to have an effect on the nano hardness and modulus properties for both the intermetallic compounds and the solder matrix. Both samples exhibited no variation in the solder with 0.3 to 0.32GPa hardness average for both. The reduced modulus was 60GPa for the FA cooled and 63GPa for the furnace cooled sample. The same can be said for IMCs with Cu6Sn5 at the joint having a hardness of 4Gpa for both samples. Therefore, hardness and modulus results values cannot justify the results found in figure 5.1 where the fracture strength declined with the furnace cooled as compared with the FA cooled sample.

76

(a)

(b) Figure 5.17 Load-Depth Curve Analysis of Furnace Cooled SnAg Solder Matrix (a) Group 2 indents (b) Group 3 indents

77

Area Contact stress vs. Creep depth for Fur. cooled SnAg 0.9 0.85 0.8 Area Contact stress (Gpa) 0.75 0.7 0.65 0.6 0.55 0.5 20 25 30 35 40 45 50 55

no trend

Creep depth (nm) (a) Creep depth vs. initial loading depth for Fur. cooled SnAg 55 50 45 creep depth (nm) 40 35 30 25 20 260 270 280 290 300 310

Initial loading depth (nm) (b) Figure 5.18 Creep Characteristics of Furnace Cooled SnAg Solder Matrix

78

Furthermore, these IMC particles also exhibited larger interactions with the nanoindenter in the form of larger projected depths of the Ag3Sn slopes found in the FA cooled than in the furnace cooled. SEM and EDX of the FA cooled sample also show these Ag3Sn micron sized precipitate particles while EDX of the furnace cooled sample shows a deficiency of them around the large Ag3Sn platelet. Further EDX evidence of the Ag 3Sn platelet formation in the furnace cooled sample is shown in figure 5.19 of another region of the furnace cooled sample. Despite the less data points in the solder matrix for creep characterization in the furnace cooled than in the FA cooled, it is sufficient to show that for a given Initial load depth, the furnace cooled sample shows less creep displacement than in the FA cooled sample and while there is a no trend in the area contact stress of the furnace cooled sample, the FA cooled exhibits this property. The creep plots highlights that there is a larger component of area contact stress with the indenter for the FA cooled sample than for the furnace cooled sample during penetration. The load-depth curve slope analysis of the furnace cooled matrix has a lot more interactions with Cu6Sn5 particles in figure 5.17 a,b (due to less Ag in the matrix but in platelets). A hardness vs. displacement profile along the joint and from substrate onto the solder matrix is shown for both samples in figure 5.20 where it can be seen that the furnace cooled sample is less conducive to stronger joints due to the numerous hard soft interfaces present in matrix.

79

Figure 5.19 SEM & EDX mapping of Furnace Cooled SnAg

80

Sn-Ag Forced Air Cooled (9.97C/s) Hardness Profile 5 4.5 4 3.5 3 Hardness (Gpa) 2.5 2 1.5 1 0.5 0 0 50 100 Displacement (m) (a) Sn-Ag Furnace Cooled (0.38C/s) Hardness Profile 5 4.5 4 3.5 3 Hardness (Gpa) 2.5 2 1.5 1 0.5 0 0 20 40 Displacement (m) (b) Figure 5.20 Hardness vs. Displacement along SnAg Sample for (a) FA cooled (b) Furnace cooled 60 80 150

Cu6Sn5
y=0 y = +20 y=0 y = +20 y=0 y = +20

Cu substrate Solder

Cu6Sn

Ag3Sn
y=0 y = +20

Cu substrate

Cu substrate

y=0

solder

81

5.2 Effect of Reflow Time on the Nanoscale Properties of 96.5Sn3.5Ag To study the effect of reflow time on Sn-Ag, two samples were selected, distinguished by their reflow times at 1s and 300s. The reflow temperature used in this study was 260C. Recalling the methodology of 3.6.2 and the results of chapter 5, the results are discussed below with less emphasis on group by group presentation but a focus on the key results. Figure 5.21 below is shows the dependence of fracture strength on the reflow time as documented by Jonathan Rowley.
FRACTURE STRENGTH OF 96.5Sn3.5Ag SOLDER WITH PEAK REFLOW TIME AT 260C
100 MEASURED VALUES 90 80 70 60 50 40 30 27.06 20 10 0 0 50 100 150 200 250 300 350 DWELL TIME (sec) 42.36 38.77 72.49 69.07 72.85 MEAN VALUES

FRACTURE STRESS (MPa)

Figures 5.21 Fracture strength of 96.5Sn3.5Ag solder joints as a function of dwell/reflow time at 260C peak reflow(23)

Figure 5.21 shows that there is degradation in mechanical integrity with longer reflow times. The sample reflowed for 1s fails at 69MPa and the 300s sample fails at 27.06 MPa. The nanoindentation study is discussed below to investigate if the nano-mechanical properties play a part in this trend.

82

5.2.1 Results for 96.5Sn3.5Ag Reflowed at 260C for 1s Four groups with13 indents/group were done on this sample. Groups1 indents are mostly on Cu with the exception of the last indent performed on Cu 6Sn5 joint and group 2 contained two indents on this IMC joint. The average hardness of the three indents is

5.70.7Gpa with an average Er of 1224Gpa. Figure 5.22 below shows the SEM and the loaddepth curves for group 2 indents. Sn-Ag (260C 1s) IMC joint

Figure 5.22 SEM of SnAg (260C 1s) Solder IMC joint Sn-Ag (260C 300s) Solder matrix (groups 3,4) Indents performed on the solder matrix for this sample recorded an average hardness of 0.33GPa and reduced modulus of 69GPa. solder matrix is shown in figure 5.23 Analysis of the load-depth curves typical of this

83

Figure 5.23 Load-Depth Curve Analysis of SnAg (260C 1s) Solder Matrix Figure 5.23 shows even interactions of Ag3Sn and Cu6Sn5 with the indenter tip, signifying a homogeneous distribution of these particles within the matrix. The EDX data in figure 5.24 from the joint to the matrix shows no islands of Cu6Sn5 and very weak signals of micron sized Ag rich regions are seen in an otherwise homogeneous matrix. Effect of this microstructure on the matrix deformation behavior can be seen in the plots of contact area stress vs. creep depth and creep depth versus initial load segment depth (figure 5.25). The minor trend in figure 5.25a shows that there is come component of contact area stress inducing the creep during load dwell.

84

Figure 5.24 EDX Mapping of SnAg (260C 1s) Solder Joint

85

Contact Area Stress vs. Creep Depth of Sn-Ag (260C - 1s) 0.95 0.9 0.85 0.8 Contact Area stress (Gpa) 0.75 0.7 0.65 0.6 0.55 0.5 30 40 50 Creep depth (nm) (a) 60 70

minor trend

creep depth vs. initial loading depth of Sn-Ag (260C - 1s) 70 65 60 55 creep depth (nm) 50 45 40 35 30 200 220 240 260 280 300 320 340

initial loading depth (nm) (b) Figure 5.25 Creep Characteristics of Sn-Ag (260C 1s) solder matrix

86

5.2.2 Results for 96.5Sn3.5Ag Reflowed at 260C for 300s Four groups of indents were performed on this sample. Figure 5.26 shows the resulting image and EDX phase across the joint where more red areas signify Ag rich and more blue signify Cu rich region. The relatively small thickness allows for easing viewing of all indents across this sample. The Cu6Sn5 IMC joints characterized here possess an average hardness of 4.8GPa with a reduced modulus of 92GPa where as the solder matrix possesses a hardness of 0.31GPa and a reduced modulus of 64GPa. Analysis of the slopes of the solder matrix typical to this sample is shown in figure 5.27 where interactions of both Cu6Sn5 and Ag3Sn are observed. The spacing between the

interactions of the IMCs are fairly large here as compared with the 1s reflow time sample implying a larger volume fraction of Sn matrix here. The phase map shows a scalloped Cu 6n5 IMC at the joint with a platelet of Ag3Sn. It seems that more Ag3Sn precipitates that would normally be in the matrix have been consumed for the formation of large Ag 3Sn platelets. EDX mapping of other regions show that formation of more platelets of Ag3Sn (Figure 5.28). The creep plots of figure 5.29 show no result of contact are stress induced creep.

87

(a)

(b) Figure 5.26 SnAg (260C 300s) (a) SEM of indents (b) EDX phase map across whole joint

88

Figure 5.27 Load-Depth Curve Analysis of SnAg (260C 300s) Solder Matrix

Figure 5.28 EDX Phase Maps of other Regions in SnAg (260C 300s)

89

Contact area stress vs. Creep depth for Sn-Ag (260C - 1s) 1.05 1 0.95 0.9 Contact stress (Gpa) 0.85 0.8 0.75 0.7 0.65 0.6 20 30 40 50 60 70

no trend

Creep Depth (nm) (a)

Creep depth vs. intiitial loading depth for Sn-Ag (260C -300s) 65 60 55 50 Creep Depth (nm) 45 40 35 30 25 20 200 220 240 260 280 300 320

Initial loading depth (b) Figure 5.29 Creep Characteristics of SnAg (260C 300s)

90

5.2.3 Discussion on the Effect of Reflow Time on 96.5Sn3.5Ag In the case of changing the reflow time two extremes of 1s and 300s were tested their effects on the nanomechanical properties Sn-Ag. It is the case here too that these variables have no considerable effect on the hardness and modulus of both the IMC joint and solder matrix. Load-depth curve slope analysis of the loading segments in the matrices showed more interaction with Cu6Sn5 in the case of 300s reflow than in the case of 1s. It seems that nano-to micron sized precipitates are being consumed for the formation of Ag 3Sn platelets during longer reflow times. The 1s reflow sample seemed to have a component of a component contact stress induced creep. An explanation for this lies in the homogeneity of the 1s solder matrix where IMC particles are embedded in a tight -Sn network without a significant amount of voids

91

CHAPTER 6 EFFECT OF COOLING RATE ON THE NANOSCALE PROPERTIES OF 95.5Sn4Ag0.5Cu (SAC405) Ternary SAC405 was tested and subject to the same methodology and analysis as the Sn-Ag in chapter. Before nanoindnetation data is discussed, a quick look at the macro-

mechanical properties as affected by the cooling rate is presents. Figure 6.1 shows the effect of cooling rate on SAC 405 fracture strength, courtesy of Jonathan Rowley.

INFLUENCE OF COOLING RATE ON 95.5Sn4.0Ag0.5Cu AT 1s PEAK REFLOW

100.00 FURNACE COOL (0.38) 90.00 FORCED AIR COOL (9.97) LIQUID NITROGEN COOL (3.32) 80.00
FRACTURE STRENGTH (MPa)

ROOM AIR COOL (2.17)

70.00 67.45 64.74 60.00 63.39

50.00 47.70

40.00

30.00

20.00 0 1 2 3 4 5 6 7 8 9 10 11 COOLING RATE (C/sec)

Figure 6.1 Influence of Cooling Rate on the Fracture Strength of SAC405 Solder Joints(23) As in the case in the previous chapter, the Forced air (FA) cooled sample yields superior fracture strength than the furnace cooled sample for SAC405.

92

6.1 Effect of Cooling Rate on the Nanoscale Properties of SAC405 This section first presents the results of the FA cooled sample followed by an analysis, then results on the furnace cooled joint is presented also with its subsequent analysis. 6.1.1 Results and Analysis for Forced Air Cooled SAC405 SAC405 Solder joint (Groups 1 and 2, 500N max load) Groups 1 and 2 are loaded at a max load of 500uN where most of the indents are on the Cu substrate. Two of the indents characterize the hardness of the Cu6Sn5 joint. The two indents record an average hardness of 4.67GPa for Cu6Sn5 IMC joint with an Er of 110GPa. Figure 6.2 show the SEM groups 1 and 2 indentation. The indents on the solder did not yield curves as 500uN max load due to the acuteness of the cube corner tip SAC 405 Solder Matrix (Group 3, 200uN max load) Group 3 indents are loaded at 200uN max load where all indents are performed on the solder solder matrix. Figure 6.3 shows the SEM of group 2 indents which are on the solder matrix. Indents 10 13 are not recorded as the void on location 10 caused faulty data from in the automation from 10 13. The load depth curves for the recorded indents are shown in figure 6.4. The multiphase particle interaction with the indenter tip is evident in this alloy also from the loading profile observed and the analysis of indents 2, 3, and 6 is displayed in figure 6.4. Figure 6.4 shows slopes that match the Cu6n5 and Ag3Sn model compounds. The low gradient slopes of Sn matrix are as is expected from the homogenous phase. The -Sn slopes are twice as steep as that of the FA cooled Sn-Ag which is most likely due to addition of 0.5%Cu to the solder chemistry. Figure 6.5 shows the Cu and Ag EDX mapping across the joint

93

Figure 6.2 SEM of Groups 1 and 2 Indents across FA Cooled SAC405 Solder Joint

Figure 6.3 SEM image of Group 3 indents on FA cooled SAC405 matrix

94

250

Load - Depth Curves for FA cooled SAC405

200

Series1 Series2

150 Load (N) 100

Series3 Series4 Series6 Series7

50

Series9

0 0 100 200 Depth (nm) (a) 300 400

(b) Figure 6.4 (a) Load-Depth Curves (b) Load-Depth Curve Analysis of FA cooled SAC405 Solder Matrix

95

Figure 6.5 EDX Mapping of FA cooled SAC405 Solder joint

96

Figure 6.6 shows the plots, characterizing the creep behavior of the solder matrix. In the case of this ternary alloy, the indenter tip seems to induce more creep with increasing contact area stress. Contact Area stress vs creep depth for FA cooled SAC405 0.85 0.8 0.75 0.7 Contact Area 0.65 stress 0.6 (GPa) 0.55 0.5 0.45 0.4 30 35 40 (a) Creep depth vs. Initial loading depth for FA cooled SAC405 70 60 50 40 Creep depth (nm) 30 20 10 0 250 270 290 310 (b) Figure 6.6 Creep characteristics of FA cooled SAC405 solder matrix 330 350 370 390 Initial loading depth (nm) 45 50 55 60 Creep depth (nm)

97

6.1.2 Results and Analysis of Furnace Cooled SAC405 Four groups of indents are performed on this sample and the results are presented below. IMC Joint (Group1 and 2) Figure 6.7 shows the SEM image across the joint of the four groups. The average hardness of the Cu6Sn5 joint is 4.3GPa with an average Er of 102.5GPa. The solder matrix has an average hardness of 0.3GPa and an average Er of 61.7GPa.

Figure 6.7 SEM image of Furnace cooled SAC405 joint

98

Load - Depth curves of Group 2 indents on Furnace cooled (0.38C/s) SAC405 250

Cu6Sn5
200

solder matrix
Series4 Series6

150 Load (N) 100

Series7 Series9 Series10 Series12

50

0 0 100 200 Depth (nm) (a) 300 400

(b) Figure 6.8 Furnace cooled SAC405 (a) Load-Depth Curves (Group 2 Indents) (b) Load-Depth Curve Analysis of Solder Matrix

99

Figure 6.8 shows the load depth curve of group2 and the slope analysis of select solder matrix indents. Indents on the solder from group 2 are representative of solder indents in other groups so group 3 is omitted to avoid redundancy. The slope analysis shows an even

interaction of both Ag3Sn and Cu6Sn5 with the indenter. Ag and Cu EDX mapping of the solder matrix in figure 6.9 a homogenous distribution of Ag3Sn without the micron sized precipitates observed in the FA cooled sample and there are no Cu6Sn5 islands either. The solder matrix of this sample seems devoid of micron sized IMC precipitates.

Figure 6.9 EDX Mapping of Furnace Cooled SAC405 Solder Joint

100

The stress and creep characteristic in the furnace cooled solder matrix are characterized in figure 6.9 below.

Contact Area stress vs. Creep depth for furnace cooled SAC 405 0.95 0.9 0.85 0.8 Contact stress 0.75 (Gpa) 0.7 0.65 0.6 0.55 0.5 30 40 50 Creep depth (nm) (a) Ctreep depth vs Initial loading depth for furnace cooled SAC 405 65 60 55 50 Creep depth (nm) 45 40 35 30 240 260 280 300 320 340 Initial loading depth (nm) (b) Figure 6.10 Creep Characteristics of furnace cooled SAC405 solder matrix 60 70

no trend

101

6.1.3 Discussion on the Effect of Cooling Rate on the Nanoscale Properties of SAC405 In the case of ternary SAC405, the nanomechanical properties of hardness and reduced modulus do not differ. SEM imaging however, shows the distinction in the morphology of both Solder joints. The IMC morphology is of a smoother kind in the FA cooled joint.

However, a multidirectional IMC scalloped morphology is seen for the furnace cooled joint. Furthermore, voids in the IMC/solder matrix interface are also visible to high degree. (Figure 6.7: bottom of group 3 area). Within the solder matrix, slopes of Ag3Sn are more than Cu6Sn5 where as both IMCs are evenly indentified in the furnace cooled sample. The EDX mapping figure (6.5 and 6.8) supports this. Actually, this solder matrix of the furnace cooled SAC405 also contains Ag3Sn platelets not characterized by the indents but are shown in the EDX phase maps of other regions in the solder (figure 6.10) where blue rich regions contain more Cu and red rich region contain more Ag. It is the case here also that Ag3Sn platelets in the furnace cooled sample leave the solder matrix with less Ag3Sn precipitates. The stress and creep characteristics of the matrix in the FA cooled sample shows a component of creep induced by contact are stress. A linear correlation in both stress and creep plots is markedly different in the FA cooled sample than in the furnace cooled. The

independence of stress with creep depth implies roughly constant creep regardless of contact stress in the furnace cooled sample; this is a poor mechanical characteristic. After 300nm

initial loading depth, creep depths significantly above the trend line signify voids and this is more so the case in the furnace cooled sample, manifesting its overall poorer mechanical integrity.

102

Figure 6.11 EDX Phase Maps of other Regions in Furnace Cooled SAC405.

103

CHAPTER 7 SUMMARY & CONCLUSION The effect of reflow process variables on the nano-mechanial properties of binary and ternary lead free solders was investigated and documented in this study. During this process, it was discovered that different phases of the solder matrix during penetration could be identified by coupling low loads with a cube corner tip indenter. The consequent oscillated loading

segment was analyzed and slopes were identified to be either that of Ag 3Sn or Cu6Sn5. Experimentation did not identify any considerable differences in the nano scale properties of hardness and reduced modulus for both Intermetlic and Solder matrices characterized. However, reflow process variables seemed to have some effect on creep behavior. It was found that the FA cooled samples on both binary and ternary solders had slightly larger components of contact (normal) stress than the furnace cooled samples. One conclusive distinction is the change of IMC formation and morphology with the difference of cooling rate and reflow times. The images observed show that slow cooling rates significantly changed the IMC formation. Longer reflow times were observed to have the same effect. In the case of the 300s sample, voids were clearly visible in the interface between the IMC scallops and the solder matrix. Though the hardness of the IMC is the same in all cases with differences due to the thickness displacement from surface to the solder underneath, the lower area/volume of contact between the IMC and solder matrix due to the voids in the furnace cooled and 300s sample leads to poor macro-mechanical properties. This is reflected in the fracture strength results. Although a compositional study was not part of the scope, it should be noted the evaluation of the flatter slopes of the Sn matrix in the load segment analysis showed that the FA

104

cooled Sn-Ag-Cu alloy had Sn slope gradients twice as large as that of the Sn-Ag FA cooled alloy. This is most like due to the addition 0.5wt% Cu. To conclude, testing the binary Sn-Ag and ternary Sn-Ag-Cu with a low load cube corner nanoindention, the following observations were made, (i) The hardness and reduced modulus of intermetallics formed in the joints do not vary with cooling rate and reflow time (ii) The hardness and reduced modulus of Solder matrix in the joints do not vary with cooling rate and reflow time. (iii) The fan air cooled solder samples exhibited slight more components of contact area stress correlated with creep depth (v) Low load cube corner indentation allowed for quantifiable interaction between indenter tip and IMC precipitate particles and subsequent load-depth curve analysis could be used to identify the IMC.

105

APPENDIX A

NANOINDENTER DATA REEULTS FOR ALL SOLDER JOINT TEST SAMPLES

106

Nanoindenter data for Forced Air Cooled 96.5Sn0.5Ag

GROUP 1 No. of indents 1 2 3 4 5 6 7 8 9 10 11 12 13 Displacement (m) 0 2.65 5.3 7.95 10.6 13.25 15.9 18.55 21.2 23.85 26.5 29.15 31.8 Pmax(N) 499.83823 499.87847 499.75776 498.03349 499.96053 499.73544 498.27878 499.89151 499.96556 499.89426 499.82778 499.83676 499.89448 hc(nm) 152.153652 162.462498 152.67954 153.64799 161.692441 162.849049 170.234122 156.71729 155.124119 150.736426 158.517932 152.315198 160.421477 Er(GPa) 118.054501 119.831918 121.529308 115.246347 127.301924 105.620782 123.727308 120.526133 115.357767 117.070561 118.652066 113.869324 121.526021 Hardness(GPa) 2.458705 2.222232 2.445313 2.413301 2.239032 2.21342 2.059799 2.349599 2.387264 2.494546 2.308211 2.454693 2.266276 Material Cu Cu Cu Cu Cu Cu Cu Cu Cu Cu Cu Cu Cu

GROUP 2 No. of indents 1 2 3 4 5 6 Displacement (m) 0 2.65 5.3 7.95 10.6 13.25 Pmax(N) 499.97492 499.85093 499.76179 500.0476 499.76927 499.80683 hc(nm) 164.840408 155.413731 155.251698 157.513133 143.629643 159.251649 Er(GPa) 128.222828 124.20609 119.810739 110.253096 114.633103 113.73751 Hardness(GPa) 2.173076 2.379867 2.383271 2.332026 2.684706 2.291691 Material Cu Cu Cu Cu Cu Cu solder/IMC layer (Cu6Sn5) solder solder solder IMC (Cu6Sn5) solder solder

7 8 9 10 11 12 13

15.9 18.55 21.2 23.85 26.5 29.15 31.8

499.87679 500.03982 -

272.926302 106.553348 -

92.220333 90.048766 -

0.97906 4.14849 -

107

GROUP 3 No. of indents 1 2 3 4 5 6 7 8 9 10 11 12 13 Displacement (m) 0 2.65 5.3 7.95 10.6 13.25 15.9 18.55 21.2 23.85 26.5 29.15 31.8 Pmax(N) 199.899142 199.934329 199.909797 200.035745 199.850926 199.872596 199.98956 199.79806 199.916538 199.895428 199.957387 199.244555 199.855742 hc(nm) 339.825868 299.806597 322.852306 319.974021 329.590512 351.631114 283.58243 268.375192 326.465582 358.128164 299.585765 247.185541 329.846331 GROUP 4 No. of indents 1 2 3 4 5 6 7 8 9 10 11 12 13 Displacement (m) 0 2.65 5.3 7.95 10.6 13.25 15.9 18.55 21.2 23.85 26.5 29.15 31.8 Pmax(N) 199.344879 199.832222 199.918405 199.973976 199.976928 199.788549 199.842055 199.930613 199.89616 199.853826 199.811222 199.897306 199.820036 hc(nm) 330.666715 310.036796 264.456641 351.180954 384.95425 338.810984 350.711968 178.827494 306.742655 359.036396 360.242618 358.45153 335.873819 47.665747 45.266998 61.746489 21.787539 46.31683 51.058294 64.358093 Er(GPa) 53.44697 55.648619 52.739867 Hardness(GPa) 0.287663 0.319504 0.411708 0.262208 0.226559 0.277351 0.262593 0.765205 0.325087 0.252985 0.251584 0.253697 0.281266 Material solder solder solder solder solder solder solder solder solder solder solder solder solder Er(GPa) 62.440692 55.220983 64.700011 68.347916 67.893957 62.498475 53.536854 53.891788 60.641682 56.971266 Hardness(GPa) 0.276187 0.337205 0.299671 0.304164 0.289893 0.261542 0.368533 0.40194 0.294419 0.25406 0.337639 0.45687 0.289542 Material solder solder solder solder solder solder solder solder solder solder solder solder solder

108

GROUP 5 No. of indents 1 2 3 4 5 6 7 8 9 10 11 12 13 Displacement (m) 0 2.65 5.3 7.95 10.6 13.25 15.9 18.55 21.2 23.85 26.5 29.15 31.8 Pmax(N) 199.928667 199.853861 199.908071 199.970106 198.899736 199.796623 200.02616 199.939101 199.959874 199.098205 199.155618 199.808395 200.015264 hc(nm) 303.823947 344.662188 349.435563 332.75873 38.025857 131.91949 327.740629 152.114642 251.421753 328.393048 321.063443 353.625811 352.205352 GROUP 6 No. of indents 1 2 3 4 5 6 7 8 9 10 11 12 13 Displacement (m) 0 2.65 5.3 7.95 10.6 13.25 15.9 18.55 21.2 23.85 26.5 29.15 31.8 Pmax(N) 200.04739 199.885497 199.957717 199.818368 199.9968 199.895555 199.93595 199.888553 199.855295 199.962407 199.893595 199.213927 199.860266 hc(nm) 273.859313 322.323141 38.433945 34.616839 325.912813 260.639001 351.02327 278.342521 359.056946 312.199536 319.552791 341.93646 294.150578 51.297774 50.113987 69.386805 73.217568 95.484222 88.560187 77.057227 49.623072 53.502654 434.32779 56.576819 Er(GPa) Hardness(GPa) 0.389691 0.300418 4.183172 4.395336 0.295333 0.421299 0.262346 0.379446 0.252964 0.316193 0.304586 0.272541 0.347456 Material solder solder IMC (Cu6Sn5) IMC (Cu6Sn5) solder solder solder solder solder solder solder solder solder 56.387688 47.081888 57.795604 43.96921 71.371859 83.763493 Er(GPa) 52.638551 49.896733 42.785644 45.578541 100.622831 Hardness(GPa) 0.330126 0.269983 0.264209 0.285683 4.182867 1.219776 0.292758 0.983888 0.446283 0.290479 0.301192 0.259114 0.26105 Material solder solder solder solder IMC (Cu6Sn5) solder solder solder solder solder solder solder solder

109

Nanoindenter data for Furnace Cooled 96.5Sn0.5Ag

GROUP 1 No. of indents 1 2 3 4 5 6 7 8 9 10 11 12 13 Displacement (m) 0 2.6 5.2 7.8 10.4 13 15.6 18.2 20.8 23.4 26 28.6 31.2 Pmax(N) 199.95545 199.829319 199.947457 199.941832 199.916203 199.960031 199.844747 199.926621 200.012709 199.845091 199.87878 199.935682 199.884079 hc(nm) 73.57228 73.678409 64.076355 72.159296 72.16395 60.838636 66.555733 123.400834 62.171069 78.338632 37.798217 39.172623 350.108321 GROUP 2 No. of indents 1 2 3 4 5 6 7 8 9 10 11 12 13 Displacement (m) 0 2.6 5.2 7.8 10.4 13 15.6 18.2 20.8 23.4 26 28.6 31.2 Pmax(N) 199.819017 199.953237 199.866323 199.915286 199.899919 199.949292 200.035528 199.900964 199.971305 199.901144 199.994706 199.91621 199.955747 hc(nm) 72.98325 72.786448 34.902574 45.217071 51.914928 211.872745 350.289235 314.809389 327.902824 67.399643 59.352453 61.539896 315.769877 Er(GPa) 108.790209 103.930324 88.277819 100.604925 69.941017 46.472111 84.345156 69.566312 70.033486 82.729828 75.311765 62.740502 Hardness(GPa) 2.628857 2.637989 4.379354 3.838453 3.520223 0.585668 0.263353 0.311935 0.292448 2.846527 3.186242 3.090244 0.310511 Material Cu Cu IMC (Cu6Sn5) IMC (Cu6Sn5) IMC (Cu6Sn5) Solder Solder Solder Solder IMC (Ag3Sn) IMC (Ag3Sn) IMC (Ag3Sn) Solder Er(GPa) 144.933395 95.063383 108.184688 125.234238 112.134253 114.184335 106.107143 124.991499 108.729432 105.626522 99.315054 94.018061 41.724693 Hardness(GPa) 2.608727 2.603153 2.983387 2.661443 2.660926 3.121088 2.879798 1.346782 3.064754 2.43665 4.215783 4.143483 0.26337 Material Cu Cu Cu Cu Cu Cu Cu Cu Cu Cu IMC (Cu6Sn5) IMC (Cu6Sn5) Solder

110

GROUP 3 No. of indents 1 2 3 4 5 6 7 8 9 10 11 12 13 Displacement (m) 0 2.6 5.2 7.8 10.4 13 15.6 18.2 20.8 23.4 26 28.6 31.2 Pmax(N) 199.813128 199.944184 199.909909 199.900377 200.053463 199.80756 199.931478 200.006393 199.934358 199.83256 199.867525 199.86479 200.01822 hc(nm) 57.232488 339.117462 314.655673 302.402224 319.706816 320.106403 307.674861 101.612041 37.937827 66.647527 64.484767 62.004643 63.151383 107.03805 112.088783 106.176574 117.440716 121.767425 Er(GPa) 77.908043 55.535873 64.983552 64.210863 56.056582 67.375166 60.175991 Hardness(GPa) 3.276931 0.277168 0.312192 0.332553 0.304596 0.303618 0.323577 1.769848 4.209385 2.8759 2.965185 2.66391 2.43665 Material IMC (Ag3Sn) solder solder solder solder solder solder solder IMC (Cu6Sn5) Cu Cu Cu Cu

Nanoindenter data for Forced Air Cooled SAC405

GROUP 1 No. of indents 1 2 3 4 5 6 7 8 9 10 11 12 13 Displacement (m) 0 2.65 5.3 7.95 10.6 13.25 15.9 18.55 21.2 23.85 26.5 29.15 31.8 Pmax(N) 499.892115 500.015617 499.933311 497.726359 499.969883 499.986934 499.865591 499.882182 499.887012 499.973881 499.835896 500.02309 499.766719 hc(nm) 174.641257 168.699004 175.872869 162.118727 157.818239 167.196205 165.704908 166.57502 157.901806 173.39738 175.455733 92.740323 453.674154 Er(GPa) 125.974769 112.436295 123.088879 119.963545 116.629759 109.975134 119.295925 134.762657 115.093378 116.48094 100.222826 115.632594 77.745391 Hardness(GPa) 1.985578 2.096383 1.964023 2.219946 2.324705 2.125686 2.155005 2.137589 2.322419 2.008226 1.970952 4.985418 0.435859 Material Cu Cu Cu Cu Cu Cu Cu Cu Cu Cu Cu IMC (Cu6Sn5) Solder

111

GROUP 2 No. of indents 1 2 3 4 5 6 Displacement (m) 0 2.65 5.3 7.95 10.6 13.25 Pmax(N) 499.834093 500.001611 499.950248 499.985396 499.832143 499.780185 hc(nm) 161.944381 160.793493 159.014925 173.365692 103.261325 212.995514 GROUP 3 No. of indents 1 2 3 4 5 6 7 8 9 Displacement (m) 0 2.65 5.3 7.95 10.6 13.25 15.9 18.55 21.2 Pmax(N) 198.891162 199.827201 200.047225 199.97268 199.052045 199.845026 199.922675 200.03126 199.912971 hc(nm) 358.455886 353.808589 335.191917 356.241937 402.678103 313.667411 344.390856 407.364066 361.698827 47.120496 73.757495 53.785439 51.655166 64.543661 60.598272 66.916718 59.032892 Er(GPa) Hardness(GPa) 0.252416 0.258926 0.282498 0.256303 0.209916 0.313657 0.270415 0.207098 0.250101 Material solder solder solder solder solder solder solder solder Solder Er(GPa) 135.25273 131.401241 125.250145 115.891283 104.760428 Hardness(GPa) 2.233069 2.258645 2.297629 2.008846 4.328096 1.451687 Material Cu Cu Cu Cu IMC (Cu6Sn5) IMC/Solder

Nanoindenter data for Furnace Cooled SAC405

GROUP 1 No. of indents 1 2 3 4 5 6 7 8 9 10 11 Displacement (m) 0 2.55 5.1 7.65 10.2 12.75 15.3 17.85 20.4 22.95 25.5 Pmax(N) 499.78957 500.0446 499.94912 497.94329 499.8606 499.87826 499.87838 499.88618 500.07693 499.96753 498.23197 hc(nm) 106.878099 156.506326 162.778795 156.118691 169.793182 185.471434 164.019278 163.122125 163.901222 177.27251 101.201313 Er(GPa) 89.020167 126.328913 116.484415 133.551397 113.891586 97.55487 115.838293 111.725905 124.142389 100.306341 107.807296 Hardness(GPa) 4.129094 2.355209 2.21585 2.3543 2.074715 1.806686 2.189579 2.208338 2.192899 1.939925 4.432822 Material Cu Cu Cu Cu Cu Cu Cu Cu Cu Cu IMC (Cu6Sn5)

112

GROUP 2 No. of indents 1 2 3 4 5 6 7 8 9 10 11 12 13 Displacement (m) 0 2.55 5.1 7.65 10.2 12.75 15.3 17.85 20.4 22.95 25.5 28.05 30.6 Pmax(N) 199.972406 199.806509 199.332979 199.905031 199.832315 200.041077 199.974573 199.861169 199.885587 199.817372 199.91033 199.864042 199.838909 hc(nm) 81.945722 76.222466 89.223835 54.533511 244.096453 329.36426 326.431827 349.735664 359.054034 313.314239 248.282153 303.670893 334.726972 GROUP 3 No. of indents 1 2 3 4 5 6 7 8 9 10 11 12 13 Displacement (m) 0 2.55 5.1 7.65 10.2 12.75 15.3 17.85 20.4 22.95 25.5 28.05 30.6 Pmax(N) 199.82314 199.78422 199.8702 199.95559 199.92446 200.02888 199.97342 198.89192 199.98842 199.90722 199.96017 199.93785 199.91255 hc(nm) 338.325673 293.87083 316.782963 330.10882 303.328706 355.672435 322.007075 360.129418 342.014346 360.532311 306.366671 132.100367 37.111991 Er(GPa) 51.936665 57.22756 58.354911 63.696888 62.649571 53.290943 62.118559 42.427287 91.034563 49.234463 54.780108 5.471081 92.863749 Hardness(GPa) 0.278033 0.34785 0.3088 0.289321 0.330977 0.257029 0.30102 0.250552 0.273502 0.251383 0.325826 1.218151 4.254084 Material Solder Solder Solder Solder Solder Solder Solder Solder Solder Solder Solder Solder IMC (Cu6Sn5) Er(GPa) 120.637764 114.353532 85.197403 93.838213 72.173756 73.359222 68.923932 68.548608 51.550822 56.283083 66.632698 59.841939 72.18108 Hardness(GPa) 2.31693 2.510451 2.08601 3.40001 0.467475 0.290486 0.294553 0.263786 0.253005 0.314177 0.455181 0.330284 0.282828 Material Cu Cu Cu IMC (Cu6Sn5) Solder Solder Solder Solder Solder Solder Solder Solder Solder

113

Nanoindenter data for SnAg reflowed at 260C for 1s

GROUP 1 No. of indents 1 2 3 4 5 6 7 8 9 10 11 12 13 Displacement (m) 0 2.55 5.1 7.65 10.2 12.75 15.3 17.85 20.4 22.95 25.5 28.05 30.6 Pmax(N) 499.871655 499.979354 499.93686 499.859995 499.757561 500.01364 499.813699 499.87217 499.961896 499.850523 500.013477 497.887373 499.900835 hc(nm) 152.743933 148.97559 145.936432 158.213646 160.870525 157.665402 171.537804 186.859129 145.522575 156.246586 159.352279 165.171428 73.112291 GROUP 2 No. of indents 1 2 3 4 5 6 7 Displacement (m) 0 2.55 5.1 7.65 10.2 12.75 15.3 Pmax(N) 499.85982 499.85052 497.67881 499.89352 497.77519 499.82585 499.85903 hc(nm) 174.863143 173.9343 168.462456 175.668529 90.720418 83.81598 293.80416 Er(GPa) 137.977756 114.862789 121.823647 112.256345 117.900662 124.692355 66.973225 Hardness(GPa) 1.981509 1.998049 2.091153 1.967444 5.102604 5.64063 0.870634 Material Cu Cu Cu Cu IMC (Cu6Sn5) IMC (Cu6Sn5) Solder Er(GPa) 129.119684 128.625069 133.462907 128.118122 115.908938 122.731551 125.977629 137.222193 122.732324 124.968676 116.374658 113.196114 125.76498 Hardness(GPa) 2.444286 2.54028 2.621282 2.315225 2.255868 2.32839 2.041688 1.785601 2.63278 2.360335 2.290399 2.157266 6.564745 Material Solder Solder Solder Solder Solder Solder Solder Solder Solder Solder Solder Solder IMC (Cu6Sn5)

114

GROUP 3 No. of indents 1 2 3 4 5 6 7 8 9 10 11 12 13 Displacement (m) 0 2.55 5.1 7.65 10.2 12.75 15.3 17.85 20.4 22.95 25.5 28.05 30.6 Pmax(N) 199.999293 199.916219 199.985315 199.953613 199.970252 199.923995 199.920217 199.956932 199.843842 200.024139 199.791911 199.939312 199.827487 hc(nm) 300.838145 298.246636 347.541312 351.901209 281.968365 289.071502 324.150532 216.039879 323.517135 393.971647 326.212977 360.503674 285.613959 84.222782 51.847602 60.935875 Er(GPa) 57.523622 51.512507 66.178296 55.709813 71.790178 81.020173 65.793964 84.376646 72.199129 Hardness(GPa) 0.335475 0.339986 0.266608 0.261328 0.37186 0.357338 0.297778 0.567883 0.298593 0.218411 0.294598 0.251456 0.364073 Material Solder Solder Solder Solder Solder Solder Solder Solder Solder Solder Solder Solder Solder

Nanoindenter data for SnAg reflowed at 260C for 300s

GROUP 1 No. of indents 1 2 3 4 5 6 7 8 Displacement (m) 0 2.55 5.1 7.65 10.2 12.75 15.3 17.85 Pmax(N) 499.91354 497.9972 499.73517 500.08492 499.73229 499.81514 499.82271 499.86988 hc(nm) 169.515655 160.544279 162.797852 168.868655 174.108266 83.149026 83.875941 95.767656 Er(GPa) 116.590377 124.015104 109.064135 111.514869 89.497107 112.699017 114.780255 101.969636 Hardness(GPa) 2.080235 2.255001 2.214499 2.093391 1.994456 5.69365 5.635845 4.782747 Material Cu Cu Cu Cu Cu/IMC interface IMC IMC IMC

115

GROUP 2 No. of indents 1 2 3 4 5 6 7 8 9 10 11 12 13 Displacement (m) 0 2.55 5.1 7.65 10.2 12.75 15.3 17.85 20.4 22.95 25.5 28.05 30.6 Pmax(N) 199.91001 199.80801 200.0269 199.85338 199.92062 199.86738 200.02723 199.96859 200.02047 199.86403 199.8523 199.77795 199.91124 hc(nm) 341.190177 316.957883 284.035929 305.704314 327.109737 320.851474 304.448498 362.464658 297.55838 280.018725 336.686005 321.099921 226.314912 GROUP 3 No. of indents 1 2 3 4 5 6 7 8 9 10 11 12 13 Displacement (m) 0 2.55 5.1 7.65 10.2 12.75 15.3 17.85 20.4 22.95 25.5 28.05 30.6 Pmax(N) 199.99631 200.02119 199.78725 199.91997 199.8885 199.96329 199.8929 199.87093 199.91529 199.97499 199.97549 199.51904 199.84922 hc(nm) 312.137734 341.042372 311.625789 321.374078 342.096555 68.723032 33.159651 25.561463 65.509473 73.110647 64.172167 67.428524 66.151109 Er(GPa) 92.407455 64.424119 56.26485 55.851122 54.390447 90.239945 94.168564 90.02074 121.644987 132.495409 102.73368 116.266871 114.702903 Hardness(GPa) 0.316347 0.274788 0.316843 0.301883 0.273261 2.794646 4.487085 5.126149 2.923546 2.626152 2.979806 2.839928 2.896321 Material solder solder solder solder solder solder IMC IMC Cu Cu Cu Cu Cu Er(GPa) 82.733604 61.706194 56.387358 56.138838 46.402586 44.098709 70.823281 89.091768 67.304059 49.44733 51.550457 51.640217 81.088014 Hardness(GPa) 0.274446 0.308433 0.367665 0.326776 0.293503 0.302587 0.32921 0.24933 0.341416 0.375791 0.280232 0.302079 0.527394 Material solder solder solder solder solder solder solder solder solder solder solder solder solder

116

REFERENCES [1] A. Rahn: 1993, . [2] K. J. Puttliz and K. A. Statler: 2004, . [3] H. Tsukamoto, Z. Dong, H. Huang, T. Nishimura, and K. Nogita: Materials Science and Engineering: B, 2009, vol. 164, 44. [4] R. R. Chromik, R. P. Vinci, S. L. Allen, and M. R. Notis: J. Mater. Res., 2003, vol. 18, 2251. [5] C. Z. Liu and J. Chen: Materials Science and Engineering A, 2007, vol. 448, 340. [6] R. R. Chromik, R. P. Vinci, S. L. Allen, and M. R. Notis: JOM, 2003, vol. 55, 66. [7] H. -. Albrecht, T. Hannach, A. Hase, A. Juritza, K. Muller, and W. H. Muller: 2004, , 462. [8] W. H. Muller, H. Worrack, J. Sterthaus, J. Villain, J. Wilden, and A. Juritza: Microsystem Technologies, 2009, vol. 15, 45. [9] B. Huang: Proceedings 1999 International Symposium on Microelectronics, 1999, vol. 3906, 711. [10] S. Ganesan and M. G. Pecht: 2006, , 766. [11] NIST, Material Science and Engineering Laboratory, Metallurgy Division: 2009, . [12] M. G. Pecht: 1993, , 296. [13] J. Bath: 2007, , 299. [14] SMTSOLDERPASTE.COM: 2008, vol. 2009, 1. [15] K. S. Kim, S. H. Huh, and K. Suganuma: Materials Science and Engineering A, 2002, vol. 333, 106. [16] S. W. Jeong, J. H. Kim, and H. M. O. Lee: J Electron Mater, 2004, vol. 33, 1530. [17] X. Deng, N. Chawla, K. K. Chawla, and M. Koopman: Acta Materialia, 2004, vol. 52, 4291. [18] A. C. Fischer-Cripps: NANOINDENTATION, Springer, New York, 2002, pp. 190. [19] X. Li and B. Bhushan: Mater Charact, 2002, vol. 48, 11.

117

[20] X. Deng, M. Koopman, N. Chawla, and K. K. Chawla: Materials Science and Engineering A, 2004, vol. 364, 240. [21] F. Gao and T. Takemoto: Mater Lett, 2006, vol. 60, 2315. [22] Y. Sun, J. Liang, Z. Xu, G. Wang, and X. Li: J. Mater. Sci. : Mater. Electron., 2008, vol. 19, 514. [23] J. G. Rowley: 2003, .

118

BIOGRAPHICAL INFORMATION Tonye Adeogba earned his Bachelor of Science degree from the State University of New York at Buffalo, Double Majoring in Mechanical and Aerospace engineering, May 2007. He then pursued a Master of Science degree in Materials Science and Engineering at the University of Texas at Arlington. Here, he first worked on Oil additives in Dr. Aswaths Tribolgy, Coatings and Lubrications Lab group and then studied Lead Free solders for his thesis to complete his Masters degree.

119